- Email: [email protected]
Epitopes in the interacting regions of β-dystroglycan (PPxY motif) and dystrophin (WW domain)
Biochimica et Biophysica Acta 1527 (2001) 54^60
www.bba-direct.com
Epitopes in the interacting regions of L-dystroglycan (PPxY motif) and dystrophin (WW domain) Aleksandr V. Pereboev 1 , Naseem Ahmed 2 , Nguyen thi Man, Glenn E. Morris * MRIC Biochemistry Group, North East Wales Institute, Mold Road, Wrexham LL11 2AW, UK Received 29 January 2001; received in revised form 20 April 2001; accepted 23 April 2001
Abstract The dystroglycan gene produces two products from a single mRNA, the extracellular K-dystroglycan and the transmembrane L-dystroglycan. The Duchenne muscular dystrophy protein, dystrophin, associates with the muscle membrane via L-dystroglycan, the WW domain of dystrophin interacting with a PPxY motif in L-dystroglycan. A panel of four monoclonal antibodies (MANDAG1^4) was produced using the last 16 amino acids of L-dystroglycan as immunogen. The mAbs recognized a 43 kDa band on Western blots of all cells and tissues tested and stained the sarcolemma in immunohistochemistry of skeletal muscle over a wide range of animal species. A monoclonal antibody (mAb) against the WW domain of dystrophin, MANHINGE4A, produced using a 16-mer synthetic peptide, recognized dystrophin on Western blots and also stained the sarcolemma. We have identified the precise sequences recognized by the mAbs using a phage-displayed random 15-mer peptide library. A 7-amino-acid consensus sequence SPPPYVP involved in binding all four L-dystroglycan mAbs was identified by sequencing 17 different peptides selected from the library. PPY were the most important residues for three mAbs, but PxxVP were essential residues for a fourth mAb, MANDAG2. By sequencing five different random peptides from the library, the epitope on dystrophin recognized by mAb MANHINGE4A was identified as PWxRA in the first L-strand of the WW domain, with the W and R residues invariably present. A recent three-dimensional structure confirms that the two epitopes are adjacent in the dystrophin^dystroglycan complex, highlighting the question of how the two interacting motifs can also be accessible to antibodies during immunolocalization in situ. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Dystrophin ; Dystroglycan; Muscular dystrophy ; Phage display peptide library ; Epitope mapping; Monoclonal antibody
1. Introduction Dystroglycan is produced as a precursor protein that is proteolytically cleaved into two interacting subunits, Kand L-dystroglycan [1]. K-Dystroglycan is a external membrane glycoprotein that interacts directly with laminins in the extracellular matrix [1,2]. L-Dystroglycan, a membrane-spanning glycoprotein, binds on the cytoplasmic side of muscle membranes to the WW and E-F hand domains of dystrophin [3], though it is also involved in signal transduction a¡ecting cytoskeletal organization [4]. Dystrophin is predominantly a muscle protein and is absent or defective in Duchenne and Becker muscular dystrophies
* Corresponding author. Fax: +44-1978-290008; E-mail : [email protected] 1 Present address: Gene Therapy Center, University of Alabama at Birmingham, 1919 Seventh Avenue South, Birmingham, AL 35294, USA. 2 Present address: Laboratory of Developmental Immunology, Harvard Medical School and Massachusetts General Hospital, Fruit Street, Boston, MA 02114-2696, USA.
[5]. Two other proteins with WW domains, utrophin [6] and caveolin [7], also interact with L-dystroglycan. The dystrophin-anchoring site on L-dystroglycan includes a PPxY motif within its last 15 amino acids [3,8] and interaction is prevented by phosphorylation of the tyrosine residue [9]. The N-terminus of dystrophin binds to cytoskeletal actin, so that the dystrophin^dystroglycan complex links the extracellular matrix to the intracellular cytoskeleton [10,11]. The loss of this link in muscular dystrophy is thought to render the sarcolemma sensitive to mechanical damage, leading to muscle degeneration and wasting [10,11]. The dystrophin^dystroglycan interaction may therefore be of fundamental importance to normal muscle function and maintenance. Epitope mapping is a useful technique for determining where antibodies bind within a large protein antigen [12]. When short peptide sequences from proteins are used as immunogens, as they increasingly are, mapping is often thought unnecessary because the binding site is already known to lie within the length of the peptide. We have already brie£y described the production of monoclonal
0304-4165 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 0 1 ) 0 0 1 4 7 - 7
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
antibodies (mAbs) against L-dystroglycan [13] and dystrophin [14] using synthetic 15-mer or 16-mer peptides conjugated to carrier proteins. We now describe the use of a phage-displayed random peptide library to identify the epitopes with great precision, since individual amino-acid residues that are important for mAb binding are identi¢ed. We show that the mAbs recognize epitopes of 5^7 amino acids within the 15/16-mer peptides and these epitopes correspond exactly to interacting surfaces in L-dystroglycan and dystrophin. These mAbs may be useful for in vitro studies of dystrophin interactions [15]. 2. Materials and methods 2.1. Monoclonal antibody production Peptides corresponding to the last 15 amino acids (KNMTPYRSPPPYVPP) of L-dystroglycan and amino acids 3054^3069 (STSVQGPWERAISPNK) were synthesized (Alta Bioscience, University of Birmingham, UK) and conjugated to BSA using glutaraldehyde. Balb/c mice were immunized and hyperimmune spleen cells isolated for hybridoma fusion with the Sp2/0 mouse myeloma cell line as described previously [16,17]. For L-dystroglycan hybridomas, over 600 out of 768 wells plated showed hybridoma growth and 69 were positive in an enzymelinked immunosorbent assay (ELISA) screen using unconjugated peptide attached to the ELISA plate. Seven of these 69 gave a 43 kDa band on Western blots of human skeletal muscle and also stained the sarcolemma of human and mouse muscle in immunohistochemistry. Four of these seven produced stable hybridoma lines after two rounds of limited dilution cloning and the mAbs produced were named MANDAG1^4. The dystrophin mAb, MANHINGE4A, was produced from a hybridoma fusion in which 169 wells out of 768 showed hybridoma growth and four were positive when screened by ELISA against the free peptide. Two of these four also recognized dystrophin on human muscle tissue sections and Western blots and MANHINGE4A was cloned twice by limiting dilution. 2.2. Epitope mapping The 15-mer peptide display phage library and K91Kan strain of Escherichia coli cells were a generous gift from G.P. Smith (University of Missouri, MO). Phage library screening was carried out essentially as described by Smith [18] but with the modi¢cations we have described in detail elsewhere [19]. mAb MANDAG2 was puri¢ed from ascitic £uid by 50% ammonium sulphate precipitation followed by dialysis against TBS (150 mM NaCl, 25 mM Tris, pH 7.4) and was diluted 1:200 with TBS for direct adsorption onto 30 mm sterile plastic dishes. Rabbit-anti (mouse Ig) antibodies (Dako, Denmark, 1 ml of 1:200 dilution in
55
TBS) adsorbed onto dishes were used to capture mAbs MANDAG3 and 4 or MANHINGE4A from culture supernatants diluted 1:50 with TBS^0.05% Tween 20. After blocking any remaining binding sites on the dishes with 4% BSA in TBS, 1011 TU of phage library were added to each plate in 500 Wl TBS^0.05% Tween 20. The phage library had been pre-incubated on dishes with adsorbed mouse serum or rabbit-anti (mouse Ig) antibodies to eliminate phage particles that bind non-speci¢cally. After the ¢rst round of biopanning, the eluates were concentrated 10-fold using Vivaspin-500 concentrators (Vivascience, UK) and were then used to infect K91Kan E. coli cells. A portion of these cells was plated on LB-agar plates containing 100 Wg/ml kanamycin and 40 Wg/ml tetracycline and the rest were grown overnight in the presence of 40 Wg/ml tetracycline to amplify the bound phage. These phages were then puri¢ed by a PEG-precipitation method [19] and used for the next biopanning step. Two rounds of biopanning were performed. Colonies of the infected cells after both rounds were grown on nitrocellulose membranes and screened with mAbs as for Western blotting to reveal positive clones [19]. Phage DNA was isolated and puri¢ed from positive clones [19] and the expressed peptide was identi¢ed by DNA sequencing using Sequenase 2.0 and the phage-speci¢c primer 5P-CCCTCATAGTTAGCGTAACG. 2.3. Western blotting Cells and tissues were homogenized and boiled for 2 min in 4^9 volumes of extraction bu¡er (2% sodium dodecyl sulphate (SDS), 5% 2-mercaptoethanol, 62.5 mM Tris^HCl, pH 6.8) before centrifugation for 1 h at 100 000Ug. Supernatants were analysed on 3^12.5% gradient polyacrylamide gels and transferred to nitrocellulose sheets (Schleicher and Schuell BA85) electrophoretically at 100 mA for 16 h in 25 mM Tris, 192 mM glycine. After blocking for 1 h in 3% skimmed milk powder in incubation bu¡er (IB, 0.05% Triton X-100 in PBS), blots were incubated with mAb (1/100 dilution of culture supernatants in IB/1% horse serum/1% foetal calf serum/0.1% BSA) for 1 h at 20³C. After washing three times for 5 min in PBS, blots were incubated with biotinylated anti-mouse Ig and a peroxidase^avidin detection reagent (Vectastain ABC Elite, Vector Laboratories, Peterborough, UK) according to the manufacturer's instructions. After washing four times in PBS, the blot was developed with 0.4 mg/ml diaminobenzidine in 25 mM phosphate^citrate bu¡er (pH 5.0) with 0.012% H2 O2 . 2.4. Subcellular fractionation Cell pellets were rinsed with PBS, harvested and homogenized in 9 volumes of RSB bu¡er (10 mM NaCl, 1.5 mM MgCl2 ; 10 mM Tris^HCl, pH 7.5) using a Dounce ho-
BBAGEN 25192 13-6-01
56
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
mogenizer. After centrifugation at 10 000Ug for 10 min, the pellet was homogenized in RSB plus 1% Triton X-100. The homogenate was centrifuged at 10 000Ug for 10 min and the pellet was resuspended in RSB. All three fractions were boiled with an equal volume of 2USDS sample bu¡er (2% SDS, 10% 2-mercaptoethanol, 125 mM Tris^HCl, pH 6.8) for Western blotting. 2.5. Immuno£uorescence microscopy Frozen sections (5 Wm thickness) of human or rabbit skeletal muscle were air-dried and incubated with mAb (1/10 dilution in PBS/1% horse serum/1% foetal calf serum) for 1 h, washed for 3U5 min with PBS and incubated with £uorescein isothiocyanate-conjugated rabbit anti-(mouse Ig) (DAKO ; 1/40 dilution) for 30 min. After washing 3U5 min with PBS, sections were mounted in 10% glycerol in PBS and photographed using a Leica DMRB £uorescence photomicroscope with 10U or 40U objectives. 3. Results The speci¢city of the mAbs was demonstrated by immuno£uorescence microscopy, Western blotting and subcellular fractionation. Fig. 1 shows that MANDAG2 gives sarcolemmal staining of both human and rabbit skeletal muscle ¢bres, as expected from the known distribution of L-dystroglycan. Similar staining has been reported previously for MANDAG1 [13] and was also seen with MANDAG3 and 4 (Fig. 1). Sarcolemmal staining was also obtained with skeletal muscle from mouse, frog (Xenopus laevis) and ¢sh (Raja clavata), showing the wide species speci¢city of the mAbs (data not shown). Fig. 2a shows that MANDAG2 stains a 43 kDa band on Western blots in a variety of cells and tissues. There are also several lower Mr bands which may be degradation products. They are rather less likely to be cross-reacting proteins
because they are recognized by all four mAbs (data not shown). Fig. 2b shows a simple fractionation of rat glioma cells into a soluble, cytosolic fraction (S), a Triton-soluble, membrane fraction (T) and an insoluble pellet (P). Most of the 43 kDa band (and the smaller 25 kDa band) is found in the membrane fraction, consistent with an integral membrane protein like L-dystroglycan. The dystrophin mAb, MANHINGE4A, recognizes 427kD dystrophin on Western blots of human skeletal muscle, but does not recognize the 71 kDa C-terminal short-form of dystrophin, Dp71 [20], in HeLa cells (Fig. 3). The sequence shared between dystrophin and Dp71 begins immediately after the MANHINGE4A peptide. Unlike the control mAb, MANDRA1, against the dystrophin C-terminus, MANHINGE4A also recognizes two lower Mr proteins in skeletal muscle (30^35 kDa; Fig. 3). The smaller of these two proteins is also present in adult and foetal brain and MANHINGE4A also crossreacts with the larger protein in rabbit and mouse muscle (data not shown). Immunolocalization with MANHINGE4A shows dystrophin at the sarcolemma (Fig. 3), but with additional patchy cytoplasmic staining which may be due to one of the lower Mr proteins on the Western blot. The staining pattern is similar to another dystrophin mAb that cross-reacts with K-actinin [21] and is consistent with a myo¢brillar localization. Although the epitopes recognized by all four dystroglycan mAbs must lie within the last 15 amino acids of L-dystroglycan, we wished to determine the exact nature of each mAb epitope and any correlation with mAb properties. In a ¢rst experiment, a random library of 15-mer peptides expressed on the surface of ¢lamentous phage was selected by bio-panning with either MANDAG1 or MANDAG2 attached to solid-phase. Individual colonies obtained after two rounds of bio-panning were grown on nitrocellulose sheets and the phage particles that they released were tested for mAb binding by incubation of the nitrocellulose with mAb, in the same manner as a Western blot [19]. Fig. 4a,b shows results from 15 clones containing
Fig. 1. All MANDAG mAbs detect L-dystroglycan at the sarcolemma of rabbit or human skeletal muscle sections. Rabbit sections were photographed with a 40U objective and human sections with a 10U objective and the bars correspond to 50 mm.
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
57
six di¡erent peptide sequences. A consensus epitope sequence of seven amino acids, SPPPYVP, near the L-dystroglycan C-terminus clearly emerges from Fig. 4a,b. Amino acids outside this sequence do not contribute to mAb binding. In a further experiment, bio-panning was performed with a mixture of MANDAG3 and MANDAG4 and a further 11 di¡erent sequences were identi¢ed that con¢rmed the same consensus epitope (Fig. 4c,d). The MANDAG2 peptides (Fig. 4b) di¡er signi¢cantly from those selected by MANDAG1 (Fig. 4a) or MANDAG3/ 4 (Fig. 4c), since xxPxxVP is always present with MANDAG2, while xxPPYxx is the most common component for the other three mAbs. Overall, the central three amino acids, PPY, of the consensus epitope are the most important for mAb binding since they were present in 15/17,
Fig. 2. Western blots of L-dystroglycan in tissues and cells using mAb MANDAG2. (a) MANDAG2 detects a 43 kDa band of L-dystroglycan (L-DG) in extracts of human skeletal muscle (Mu), brain (Br), HeLa cells and primary human myoblast cultures (Myo). Pre-stained Mr markers (Sigma) were used. (b) Rat C6 glioma cells were fractionated in soluble (S, cytosolic), Triton-X-100 extractable (T, membrane-bound) and residual pellet (P). The 43 kDa L-dystroglycan was detected mainly in the membrane-bound fraction by MANDAG2.
Fig. 3. Characterization of MANHINGE4A mAb against dystrophin. On Western blots of human skeletal muscle, MANHINGE4A recognizes mainly dystrophin (DYS ; 427 kDa) and two low Mr bands, whereas the C-terminal mAb, MANDRA1 [28], recognizes only dystrophin and a non-speci¢c band (this blot was overloaded to reveal cross-reacting bands). In HeLa cells, MANHINGE4A does not detect the Dp71 band detected by MANDRA1 (two other bands present with both mAbs are non-speci¢c cross-reactions of the secondary anti-mouse Ig antibody). On human skeletal muscle sections, MANHINGE4A shows cytoplasmic staining in addition to typical dystrophin staining at the sarcolemma (cf. Fig. 1). Total width of photograph = 75 Wm.
BBAGEN 25192 13-6-01
58
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
Fig. 4. Epitope mapping of four mAbs against L-dystroglycan using a phage-displayed 15-mer peptide library. Three selection experiments (a^c) are described in the text and the original peptide immunogen is shown in (d). Residues in selected peptides that match the original sequence are shown in boldface and underlined.
16/17 and 12/17, respectively and the last ¢gure increases to 17/17 if we allow replacement of Y by other aromatic side-chains, F or W. The ¢rst two amino acids of the consensus, SP, were present in 4/17 and 5/17 peptides, while the last two, VP, were essential for MANDAG2 and present overall in 8/17 and 9/17 (though if we allow replacement of V by other hydrophobic aliphatics, L or A, the 8/17 increases to 14/17). Fig. 5 shows ¢ve peptides selected from the same library by MANHINGE4A mAb against dystrophin. These clearly show that WxRA are the most important residues for the epitope, though the three preceding amino acids may make some contribution. 4. Discussion We have shown that four mAbs prepared against a 15amino-acid peptide at the C-terminus of L-dystroglycan recognize epitopes that are largely contained within a linear sequence of seven amino acids SPPPYVP within the peptide. When some peptides selected by antibody have only three of the seven consensus residues (Fig. 4), the replacement residues must be ones that are well tolerated. James et al. [9] have also mapped the MANDAG2 epitope using a di¡erent method, in which the e¡ect of amino acid substitution was studied in chemically-synthesized 15-mer peptides. They found that, in four residues of PxYVP, the only permitted substitution without loss of MANDAG2 binding was Y892 to F (in agreement with Fig. 4b), but substitution of P891 within this sequence had no e¡ect. Our results identify the same sequence but suggest that P891 is an important residue for MANDAG2 binding because it was present in three out of four peptides selected from the random peptide library. The likely explanation for the apparent con£ict is that P891 becomes less important as the number of other residues matching the epitope increases, so that it is not needed at all when the
other 14 residues in the screening peptide match the dystroglycan sequence. All `functional' approaches to epitope mapping tend to identify only the most important residues for binding, while `structural' methods, like X-ray crystallography, tend to reveal a larger number of `contact' residues [22,23]. The ¢t between antibody and antigen, like all protein-protein interactions, is determined by the shape, charge, hydrophilicity and hydrogen bonding of amino acid side-chains at the interface. It seems likely that the four dystroglycan mAbs recognize a linear epitope containing at least SPPPYVP. MANDAG2 selected one peptide from the library with this exact sequence of seven amino acids (Fig. 4b), suggesting that all these residues contribute to the binding a¤nity. The other three mAbs may ¢t the antigen less exactly outside the PPY core of their epitopes and be more tolerant of substitutions. The epitope consensus contains the PPxY motif that is known to be involved in the interaction of L-dystroglycan with dystrophin. The three-dimensional (3D) structure of a complex between this 15-mer L-dystroglycan peptide and a sub-domain of dystrophin has recently been determined [23] and shows that the L-dystroglycan peptide lies on the surface of dystrophin at a region formed by the WW domain and one of the `EF hands' of dystrophin (Fig. 6).
Fig. 5. Epitope mapping of MANHINGE4A mAb against dystrophin using a phage-displayed 15-mer peptide library. One selection experiment (a) identi¢ed ¢ve clones displaying di¡erent peptide sequences. Residues that match the original sequence (b) are shown in boldface and underlined.
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
Fig. 6. Position of the MANDAG and MANHINGE4A epitopes in the 3D structure of a dystrophin-L-dystroglycan complex. The structure of a C-terminal L-dystroglycan peptide complexed with a C-terminal dystrophin fragment ([24]; pdb accession 1EG4) was downloaded from http:// www.ncbi.nlm.nih.gov/entrez and rotated using the Cn3D 3.0 viewing program from the same website. Residue positions were identi¢ed using Cn3D and labelled in white using Adobe Photoshop. The dystroglycan peptide is at top right and the WW domain is the L-sheet below it.
The L-dystroglycan sequence is accessible on the surface of the complex and is essentially linear. The core residues of the MANHINGE4A epitope, PWxRA, form one strand of the three-stranded anti-parallel L-sheet that constitutes the WW domain (Fig. 6). This structure raises the interesting question of how antibody binding can be accommodated within it. When dystrophin and L-dystroglycan co-localize at the sarcolemma in muscle tissue sections, the PPxY motif and WW domain are interacting but the immunohistochemistry (Fig. 1) suggests that epitope recognition by the mAbs is not prevented by the interaction. The possibility that the mAbs are recognizing side-chains projecting from one accessible face is not supported by the 3D structure, which shows the side-chains of both PWxRA in dystrophin and PPPYV in dystroglycan projecting in different directions from the main chain. Linear epitopes have to become partially buried in the mAb for fully effective interaction of the mAb with their side-chains to occur. Accessibility is not a problem for Western blots, since the antigen will be disrupted and epitopes exposed by SDS treatment, but some disruption of the dystrophin^ dystroglycan structure could also required for mAb binding to occur in situ at the sarcolemma, as observed immunohistochemically. There are three possible hypotheses. One is that the mAbs do recognize the structure in Fig. 6, but only weakly since not all the side-chains of the epitopes are accessible; this is unconvincing, but cannot be ruled out. A second is that, during immunohistochemical procedures, L-dystroglycan mAbs could displace the dystrophin in situ and vice-versa. This might explain why the mAbs only recognize L-dystroglycan or dystrophin in un¢xed tissue sections (results not shown), since formalin cross-linking would prevent displacement. A third hypothesis is that the interaction in muscle is less than 100%, so that dystrophin molecules with free WW-domains and
59
L-dystroglycan molecules with free PPxY motifs are always available at the sarcolemma in vivo. This scenario has not been ruled out experimentally since dystrophin can be retained at the sarcolemma without the dystroglycan interaction; for example, by its N-terminal interactions with sub-sarcolemmal actin [25]. Finally, the four dystrophin `hinges' were originally identi¢ed as £exible regions by their susceptibility to proteolysis, both in vitro [26] and in situ [27]. Although the exact residues susceptible to proteolysis have not been identi¢ed, hinge 4 peptide sequences were chosen as likely immunogens because of their predicted £exibility [14]. It is a little surprising that the MANHINGE4 mAb obtained in this way turns out to recognize part of a L-sheet structure rather than the random coil sequence that precedes it. Acknowledgements This work was supported by grants from the Muscular Dystrophy Campaign and the Higher Education Funding Council (Wales) DevR scheme.
References [1] O. Ibraghimov-Beskrovnaya, J.M. Ervasti, C.J. Leveille, C.A. Slaughter, S.W. Sernett, K.P. Campbell, Primary structure of dystrophin-associated glycoproteins linking dystrophin to the extracellular matrix, Nature 355 (1992) 696^702. [2] J.F. Talts, Z. Andac, W. Gohring, A. Brancaccio, R. Timpl, Binding of the G domains of laminin alpha1 and alpha2 chains and perlecan to heparin, sulfatides, alpha-dystroglycan and several extracellular matrix proteins, EMBO J. 18 (1999) 863^870. [3] S. Rentschler, H. Linn, K. Deininger, M.T. Bedford, X. Espanel, M. Sudol, The WW domain of dystrophin requires EF-hands region to interact with beta-dystroglycan, J. Biol. Chem. 380 (1999) 431^442. [4] B. Yang, D. Jung, D. Motto, J. Meyer, G. Koretzky, K.P. Campbell, SH3 domain-mediated interaction of dystroglycan and Grb2, J. Biol. Chem. 270 (1995) 11711^11714. [5] E.P. Ho¡man, R.H. Brown Jr., L.M. Kunkel, Dystrophin: the protein product of the Duchenne muscular dystrophy locus, Cell 51 (1987) 919^928. [6] M. James, Nguyen thi Man, C.J. Wise, G.E. Jones, G.E. Morris, Utrophin-dystroglycan complex in membranes of adherent cultured cells, Cell Motil. Cytoskeleton 33 (1996) 163^174. [7] F. Sotgia, J.K. Lee, K. Das, M. Bedford, T.C. Petrucci, P. Macioce, M. Sargiacomo, F.D. Bricarelli, C. Minetti, M. Sudol, M.P. Lisanti, Caveolin-3 directly interacts with the C-terminal tail of beta-dystroglycan. Identi¢cation of a central WW-like domain within caveolin family members, J. Biol. Chem. 275 (2000) 38048^38058. [8] D. Jung, B. Yang, J. Meyer, J.S. Chamberlain, K.P. Campbell, Identi¢cation and characterization of the dystrophin anchoring site on beta-dystroglycan, J. Biol. Chem. 270 (1995) 27305^27310. [9] M. James, A. Nuttall, J.L. Ilsley, K. Ottersbach, J.M. Tinsley, M. Sudol, S.J. Winder, Adhesion-dependent tyrosine phosphorylation of (beta)-dystroglycan regulates its interaction with utrophin, J. Cell Sci. 113 (2000) 1717^1726. [10] J.M. Ervasti, K.P. Campbell, A role for the dystrophin-glycoprotein complex as a transmembrane linker between laminin and actin, J. Cell Biol. 122 (1993) 809^823.
BBAGEN 25192 13-6-01
60
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
[11] S.J. Winder, The membrane^cytoskeleton interface: the role of dystrophin and utrophin, J. Muscle Res. Cell Motil. 18 (1997) 617^629. [12] G.E. Morris, Epitope mapping, in: R. Rapley, J.M. Walker (Eds.), Molecular Biomethods Handbook, Humana Press, Totowa, NJ, 1998, pp. 619^630. [13] T.R. Helliwell, Nguyen thi Man, G.E. Morris, Expression of the 43 kD dystrophin-associated glycoprotein in human neuromuscular disease, Neuromusc. Disord. 4 (1994) 101^113. [14] N. Ahmed, Nguyen thi Man, G.E. Morris, Flexible hinges in dystrophin, Biochem. Soc. Trans 26 (1998) S310. [15] G.E. Morris, Nguyen thi Man, A. Pereboev, J. Kendrick-Jones, S.J. Winder, Disruption of the utrophin-actin interaction by monoclonal antibodies and prediction of an actin-binding surface of utrophin, Biochem. J. 337 (1999) 119^123. [16] Nguyen thi Man, G.E. Morris, A rapid method for generating large numbers of high-a¤nity monoclonal antibodies from a single mouse, in: J.M. Walker (Ed.), The Protein Protocols Handbook, Humana Press, Totowa, NJ, 1996, pp. 783^792. [17] Nguyen thi Man, G.E. Morris, Production of antibodies by the hybridoma method, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 377^389. [18] G.P. Smith, Cloning in fUSE vectors, Division of Biological Sciences, University of Missouri at Columbia, USA, 1992 (available directly from Prof. G.P. Smith). [19] A. Pereboev, G.E. Morris, Reiterative screening of phage display peptide libraries, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 195^206. [20] D. Lederfein, Z. Levy, N. Augier, J. Leger, G.E. Morris, O. Fuchs, D. Ya¡e, U. Nudel, 71 kD protein is a major product of the Du-
[21]
[22] [23]
[24]
[25]
[26]
[27]
[28]
chenne Muscular Dystrophy gene in brain and other non-muscle tissues, Proc. Natl. Acad. Sci. USA 89 (1992) 5346^5350. Nguyen thi Man, J.M. Ellis, I.B. Ginjaar, M.M.B. van Paassen, G.J.B. van Ommen, A.F.M. Moorman, A.J. Cartwright, G.E. Morris, Monoclonal antibody evidence for structural similarities between the central rod regions of actinin and dystrophin, FEBS Lett. 272 (1990) 109^112. van Regenmortel, Structural and functional approaches to the study of protein antigenicity, Immunol. Today 10 (1989) 266^272. F.A. Saul, P.M. Alzari, Crystallographic studies of antigen-antibody interactions, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 11^23. X. Huang, F. Poy, R. Zhang, A. Joachimiak, M. Sudol, M.J. Eck, Structure of a WW domain containing fragment of dystrophin in complex with beta-dystroglycan, Nat. Struct. Biol. 7 (2000) 634^638. T.R. Helliwell, J.M. Ellis, R.C. Mountford, R.E. Appleton, G.E. Morris, A truncated dystrophin lacking the C-terminal domains is localized at the muscle membrane, Am. J. Hum. Genet. 50 (1992) 508^514. M. Koenig, L.M. Kunkel, Detailed analysis of the repeat domain of dystrophin reveals four potential hinge segments that may confer £exibility, J. Biol. Chem. 265 (1990) 4560^4566. S. Hori, S. Ohtani, Nguyen thi Man, G.E. Morris, The N-terminal half of dystrophin is protected from proteolysis in situ, Biochem. Biophys. Res. Commun. 209 (1995) 1062^1067. G.E. Morris, S.G. Sedgwick, J.M. Ellis, A. Pereboev, J.S. Chamberlain, Nguyen thi Man, An epitope structure for the C-terminal domain of dystrophin and utrophin, Biochemistry 37 (1998) 11117^ 11127.
BBAGEN 25192 13-6-01
www.bba-direct.com
Epitopes in the interacting regions of L-dystroglycan (PPxY motif) and dystrophin (WW domain) Aleksandr V. Pereboev 1 , Naseem Ahmed 2 , Nguyen thi Man, Glenn E. Morris * MRIC Biochemistry Group, North East Wales Institute, Mold Road, Wrexham LL11 2AW, UK Received 29 January 2001; received in revised form 20 April 2001; accepted 23 April 2001
Abstract The dystroglycan gene produces two products from a single mRNA, the extracellular K-dystroglycan and the transmembrane L-dystroglycan. The Duchenne muscular dystrophy protein, dystrophin, associates with the muscle membrane via L-dystroglycan, the WW domain of dystrophin interacting with a PPxY motif in L-dystroglycan. A panel of four monoclonal antibodies (MANDAG1^4) was produced using the last 16 amino acids of L-dystroglycan as immunogen. The mAbs recognized a 43 kDa band on Western blots of all cells and tissues tested and stained the sarcolemma in immunohistochemistry of skeletal muscle over a wide range of animal species. A monoclonal antibody (mAb) against the WW domain of dystrophin, MANHINGE4A, produced using a 16-mer synthetic peptide, recognized dystrophin on Western blots and also stained the sarcolemma. We have identified the precise sequences recognized by the mAbs using a phage-displayed random 15-mer peptide library. A 7-amino-acid consensus sequence SPPPYVP involved in binding all four L-dystroglycan mAbs was identified by sequencing 17 different peptides selected from the library. PPY were the most important residues for three mAbs, but PxxVP were essential residues for a fourth mAb, MANDAG2. By sequencing five different random peptides from the library, the epitope on dystrophin recognized by mAb MANHINGE4A was identified as PWxRA in the first L-strand of the WW domain, with the W and R residues invariably present. A recent three-dimensional structure confirms that the two epitopes are adjacent in the dystrophin^dystroglycan complex, highlighting the question of how the two interacting motifs can also be accessible to antibodies during immunolocalization in situ. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Dystrophin ; Dystroglycan; Muscular dystrophy ; Phage display peptide library ; Epitope mapping; Monoclonal antibody
1. Introduction Dystroglycan is produced as a precursor protein that is proteolytically cleaved into two interacting subunits, Kand L-dystroglycan [1]. K-Dystroglycan is a external membrane glycoprotein that interacts directly with laminins in the extracellular matrix [1,2]. L-Dystroglycan, a membrane-spanning glycoprotein, binds on the cytoplasmic side of muscle membranes to the WW and E-F hand domains of dystrophin [3], though it is also involved in signal transduction a¡ecting cytoskeletal organization [4]. Dystrophin is predominantly a muscle protein and is absent or defective in Duchenne and Becker muscular dystrophies
* Corresponding author. Fax: +44-1978-290008; E-mail : [email protected] 1 Present address: Gene Therapy Center, University of Alabama at Birmingham, 1919 Seventh Avenue South, Birmingham, AL 35294, USA. 2 Present address: Laboratory of Developmental Immunology, Harvard Medical School and Massachusetts General Hospital, Fruit Street, Boston, MA 02114-2696, USA.
[5]. Two other proteins with WW domains, utrophin [6] and caveolin [7], also interact with L-dystroglycan. The dystrophin-anchoring site on L-dystroglycan includes a PPxY motif within its last 15 amino acids [3,8] and interaction is prevented by phosphorylation of the tyrosine residue [9]. The N-terminus of dystrophin binds to cytoskeletal actin, so that the dystrophin^dystroglycan complex links the extracellular matrix to the intracellular cytoskeleton [10,11]. The loss of this link in muscular dystrophy is thought to render the sarcolemma sensitive to mechanical damage, leading to muscle degeneration and wasting [10,11]. The dystrophin^dystroglycan interaction may therefore be of fundamental importance to normal muscle function and maintenance. Epitope mapping is a useful technique for determining where antibodies bind within a large protein antigen [12]. When short peptide sequences from proteins are used as immunogens, as they increasingly are, mapping is often thought unnecessary because the binding site is already known to lie within the length of the peptide. We have already brie£y described the production of monoclonal
0304-4165 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 0 1 ) 0 0 1 4 7 - 7
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
antibodies (mAbs) against L-dystroglycan [13] and dystrophin [14] using synthetic 15-mer or 16-mer peptides conjugated to carrier proteins. We now describe the use of a phage-displayed random peptide library to identify the epitopes with great precision, since individual amino-acid residues that are important for mAb binding are identi¢ed. We show that the mAbs recognize epitopes of 5^7 amino acids within the 15/16-mer peptides and these epitopes correspond exactly to interacting surfaces in L-dystroglycan and dystrophin. These mAbs may be useful for in vitro studies of dystrophin interactions [15]. 2. Materials and methods 2.1. Monoclonal antibody production Peptides corresponding to the last 15 amino acids (KNMTPYRSPPPYVPP) of L-dystroglycan and amino acids 3054^3069 (STSVQGPWERAISPNK) were synthesized (Alta Bioscience, University of Birmingham, UK) and conjugated to BSA using glutaraldehyde. Balb/c mice were immunized and hyperimmune spleen cells isolated for hybridoma fusion with the Sp2/0 mouse myeloma cell line as described previously [16,17]. For L-dystroglycan hybridomas, over 600 out of 768 wells plated showed hybridoma growth and 69 were positive in an enzymelinked immunosorbent assay (ELISA) screen using unconjugated peptide attached to the ELISA plate. Seven of these 69 gave a 43 kDa band on Western blots of human skeletal muscle and also stained the sarcolemma of human and mouse muscle in immunohistochemistry. Four of these seven produced stable hybridoma lines after two rounds of limited dilution cloning and the mAbs produced were named MANDAG1^4. The dystrophin mAb, MANHINGE4A, was produced from a hybridoma fusion in which 169 wells out of 768 showed hybridoma growth and four were positive when screened by ELISA against the free peptide. Two of these four also recognized dystrophin on human muscle tissue sections and Western blots and MANHINGE4A was cloned twice by limiting dilution. 2.2. Epitope mapping The 15-mer peptide display phage library and K91Kan strain of Escherichia coli cells were a generous gift from G.P. Smith (University of Missouri, MO). Phage library screening was carried out essentially as described by Smith [18] but with the modi¢cations we have described in detail elsewhere [19]. mAb MANDAG2 was puri¢ed from ascitic £uid by 50% ammonium sulphate precipitation followed by dialysis against TBS (150 mM NaCl, 25 mM Tris, pH 7.4) and was diluted 1:200 with TBS for direct adsorption onto 30 mm sterile plastic dishes. Rabbit-anti (mouse Ig) antibodies (Dako, Denmark, 1 ml of 1:200 dilution in
55
TBS) adsorbed onto dishes were used to capture mAbs MANDAG3 and 4 or MANHINGE4A from culture supernatants diluted 1:50 with TBS^0.05% Tween 20. After blocking any remaining binding sites on the dishes with 4% BSA in TBS, 1011 TU of phage library were added to each plate in 500 Wl TBS^0.05% Tween 20. The phage library had been pre-incubated on dishes with adsorbed mouse serum or rabbit-anti (mouse Ig) antibodies to eliminate phage particles that bind non-speci¢cally. After the ¢rst round of biopanning, the eluates were concentrated 10-fold using Vivaspin-500 concentrators (Vivascience, UK) and were then used to infect K91Kan E. coli cells. A portion of these cells was plated on LB-agar plates containing 100 Wg/ml kanamycin and 40 Wg/ml tetracycline and the rest were grown overnight in the presence of 40 Wg/ml tetracycline to amplify the bound phage. These phages were then puri¢ed by a PEG-precipitation method [19] and used for the next biopanning step. Two rounds of biopanning were performed. Colonies of the infected cells after both rounds were grown on nitrocellulose membranes and screened with mAbs as for Western blotting to reveal positive clones [19]. Phage DNA was isolated and puri¢ed from positive clones [19] and the expressed peptide was identi¢ed by DNA sequencing using Sequenase 2.0 and the phage-speci¢c primer 5P-CCCTCATAGTTAGCGTAACG. 2.3. Western blotting Cells and tissues were homogenized and boiled for 2 min in 4^9 volumes of extraction bu¡er (2% sodium dodecyl sulphate (SDS), 5% 2-mercaptoethanol, 62.5 mM Tris^HCl, pH 6.8) before centrifugation for 1 h at 100 000Ug. Supernatants were analysed on 3^12.5% gradient polyacrylamide gels and transferred to nitrocellulose sheets (Schleicher and Schuell BA85) electrophoretically at 100 mA for 16 h in 25 mM Tris, 192 mM glycine. After blocking for 1 h in 3% skimmed milk powder in incubation bu¡er (IB, 0.05% Triton X-100 in PBS), blots were incubated with mAb (1/100 dilution of culture supernatants in IB/1% horse serum/1% foetal calf serum/0.1% BSA) for 1 h at 20³C. After washing three times for 5 min in PBS, blots were incubated with biotinylated anti-mouse Ig and a peroxidase^avidin detection reagent (Vectastain ABC Elite, Vector Laboratories, Peterborough, UK) according to the manufacturer's instructions. After washing four times in PBS, the blot was developed with 0.4 mg/ml diaminobenzidine in 25 mM phosphate^citrate bu¡er (pH 5.0) with 0.012% H2 O2 . 2.4. Subcellular fractionation Cell pellets were rinsed with PBS, harvested and homogenized in 9 volumes of RSB bu¡er (10 mM NaCl, 1.5 mM MgCl2 ; 10 mM Tris^HCl, pH 7.5) using a Dounce ho-
BBAGEN 25192 13-6-01
56
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
mogenizer. After centrifugation at 10 000Ug for 10 min, the pellet was homogenized in RSB plus 1% Triton X-100. The homogenate was centrifuged at 10 000Ug for 10 min and the pellet was resuspended in RSB. All three fractions were boiled with an equal volume of 2USDS sample bu¡er (2% SDS, 10% 2-mercaptoethanol, 125 mM Tris^HCl, pH 6.8) for Western blotting. 2.5. Immuno£uorescence microscopy Frozen sections (5 Wm thickness) of human or rabbit skeletal muscle were air-dried and incubated with mAb (1/10 dilution in PBS/1% horse serum/1% foetal calf serum) for 1 h, washed for 3U5 min with PBS and incubated with £uorescein isothiocyanate-conjugated rabbit anti-(mouse Ig) (DAKO ; 1/40 dilution) for 30 min. After washing 3U5 min with PBS, sections were mounted in 10% glycerol in PBS and photographed using a Leica DMRB £uorescence photomicroscope with 10U or 40U objectives. 3. Results The speci¢city of the mAbs was demonstrated by immuno£uorescence microscopy, Western blotting and subcellular fractionation. Fig. 1 shows that MANDAG2 gives sarcolemmal staining of both human and rabbit skeletal muscle ¢bres, as expected from the known distribution of L-dystroglycan. Similar staining has been reported previously for MANDAG1 [13] and was also seen with MANDAG3 and 4 (Fig. 1). Sarcolemmal staining was also obtained with skeletal muscle from mouse, frog (Xenopus laevis) and ¢sh (Raja clavata), showing the wide species speci¢city of the mAbs (data not shown). Fig. 2a shows that MANDAG2 stains a 43 kDa band on Western blots in a variety of cells and tissues. There are also several lower Mr bands which may be degradation products. They are rather less likely to be cross-reacting proteins
because they are recognized by all four mAbs (data not shown). Fig. 2b shows a simple fractionation of rat glioma cells into a soluble, cytosolic fraction (S), a Triton-soluble, membrane fraction (T) and an insoluble pellet (P). Most of the 43 kDa band (and the smaller 25 kDa band) is found in the membrane fraction, consistent with an integral membrane protein like L-dystroglycan. The dystrophin mAb, MANHINGE4A, recognizes 427kD dystrophin on Western blots of human skeletal muscle, but does not recognize the 71 kDa C-terminal short-form of dystrophin, Dp71 [20], in HeLa cells (Fig. 3). The sequence shared between dystrophin and Dp71 begins immediately after the MANHINGE4A peptide. Unlike the control mAb, MANDRA1, against the dystrophin C-terminus, MANHINGE4A also recognizes two lower Mr proteins in skeletal muscle (30^35 kDa; Fig. 3). The smaller of these two proteins is also present in adult and foetal brain and MANHINGE4A also crossreacts with the larger protein in rabbit and mouse muscle (data not shown). Immunolocalization with MANHINGE4A shows dystrophin at the sarcolemma (Fig. 3), but with additional patchy cytoplasmic staining which may be due to one of the lower Mr proteins on the Western blot. The staining pattern is similar to another dystrophin mAb that cross-reacts with K-actinin [21] and is consistent with a myo¢brillar localization. Although the epitopes recognized by all four dystroglycan mAbs must lie within the last 15 amino acids of L-dystroglycan, we wished to determine the exact nature of each mAb epitope and any correlation with mAb properties. In a ¢rst experiment, a random library of 15-mer peptides expressed on the surface of ¢lamentous phage was selected by bio-panning with either MANDAG1 or MANDAG2 attached to solid-phase. Individual colonies obtained after two rounds of bio-panning were grown on nitrocellulose sheets and the phage particles that they released were tested for mAb binding by incubation of the nitrocellulose with mAb, in the same manner as a Western blot [19]. Fig. 4a,b shows results from 15 clones containing
Fig. 1. All MANDAG mAbs detect L-dystroglycan at the sarcolemma of rabbit or human skeletal muscle sections. Rabbit sections were photographed with a 40U objective and human sections with a 10U objective and the bars correspond to 50 mm.
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
57
six di¡erent peptide sequences. A consensus epitope sequence of seven amino acids, SPPPYVP, near the L-dystroglycan C-terminus clearly emerges from Fig. 4a,b. Amino acids outside this sequence do not contribute to mAb binding. In a further experiment, bio-panning was performed with a mixture of MANDAG3 and MANDAG4 and a further 11 di¡erent sequences were identi¢ed that con¢rmed the same consensus epitope (Fig. 4c,d). The MANDAG2 peptides (Fig. 4b) di¡er signi¢cantly from those selected by MANDAG1 (Fig. 4a) or MANDAG3/ 4 (Fig. 4c), since xxPxxVP is always present with MANDAG2, while xxPPYxx is the most common component for the other three mAbs. Overall, the central three amino acids, PPY, of the consensus epitope are the most important for mAb binding since they were present in 15/17,
Fig. 2. Western blots of L-dystroglycan in tissues and cells using mAb MANDAG2. (a) MANDAG2 detects a 43 kDa band of L-dystroglycan (L-DG) in extracts of human skeletal muscle (Mu), brain (Br), HeLa cells and primary human myoblast cultures (Myo). Pre-stained Mr markers (Sigma) were used. (b) Rat C6 glioma cells were fractionated in soluble (S, cytosolic), Triton-X-100 extractable (T, membrane-bound) and residual pellet (P). The 43 kDa L-dystroglycan was detected mainly in the membrane-bound fraction by MANDAG2.
Fig. 3. Characterization of MANHINGE4A mAb against dystrophin. On Western blots of human skeletal muscle, MANHINGE4A recognizes mainly dystrophin (DYS ; 427 kDa) and two low Mr bands, whereas the C-terminal mAb, MANDRA1 [28], recognizes only dystrophin and a non-speci¢c band (this blot was overloaded to reveal cross-reacting bands). In HeLa cells, MANHINGE4A does not detect the Dp71 band detected by MANDRA1 (two other bands present with both mAbs are non-speci¢c cross-reactions of the secondary anti-mouse Ig antibody). On human skeletal muscle sections, MANHINGE4A shows cytoplasmic staining in addition to typical dystrophin staining at the sarcolemma (cf. Fig. 1). Total width of photograph = 75 Wm.
BBAGEN 25192 13-6-01
58
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
Fig. 4. Epitope mapping of four mAbs against L-dystroglycan using a phage-displayed 15-mer peptide library. Three selection experiments (a^c) are described in the text and the original peptide immunogen is shown in (d). Residues in selected peptides that match the original sequence are shown in boldface and underlined.
16/17 and 12/17, respectively and the last ¢gure increases to 17/17 if we allow replacement of Y by other aromatic side-chains, F or W. The ¢rst two amino acids of the consensus, SP, were present in 4/17 and 5/17 peptides, while the last two, VP, were essential for MANDAG2 and present overall in 8/17 and 9/17 (though if we allow replacement of V by other hydrophobic aliphatics, L or A, the 8/17 increases to 14/17). Fig. 5 shows ¢ve peptides selected from the same library by MANHINGE4A mAb against dystrophin. These clearly show that WxRA are the most important residues for the epitope, though the three preceding amino acids may make some contribution. 4. Discussion We have shown that four mAbs prepared against a 15amino-acid peptide at the C-terminus of L-dystroglycan recognize epitopes that are largely contained within a linear sequence of seven amino acids SPPPYVP within the peptide. When some peptides selected by antibody have only three of the seven consensus residues (Fig. 4), the replacement residues must be ones that are well tolerated. James et al. [9] have also mapped the MANDAG2 epitope using a di¡erent method, in which the e¡ect of amino acid substitution was studied in chemically-synthesized 15-mer peptides. They found that, in four residues of PxYVP, the only permitted substitution without loss of MANDAG2 binding was Y892 to F (in agreement with Fig. 4b), but substitution of P891 within this sequence had no e¡ect. Our results identify the same sequence but suggest that P891 is an important residue for MANDAG2 binding because it was present in three out of four peptides selected from the random peptide library. The likely explanation for the apparent con£ict is that P891 becomes less important as the number of other residues matching the epitope increases, so that it is not needed at all when the
other 14 residues in the screening peptide match the dystroglycan sequence. All `functional' approaches to epitope mapping tend to identify only the most important residues for binding, while `structural' methods, like X-ray crystallography, tend to reveal a larger number of `contact' residues [22,23]. The ¢t between antibody and antigen, like all protein-protein interactions, is determined by the shape, charge, hydrophilicity and hydrogen bonding of amino acid side-chains at the interface. It seems likely that the four dystroglycan mAbs recognize a linear epitope containing at least SPPPYVP. MANDAG2 selected one peptide from the library with this exact sequence of seven amino acids (Fig. 4b), suggesting that all these residues contribute to the binding a¤nity. The other three mAbs may ¢t the antigen less exactly outside the PPY core of their epitopes and be more tolerant of substitutions. The epitope consensus contains the PPxY motif that is known to be involved in the interaction of L-dystroglycan with dystrophin. The three-dimensional (3D) structure of a complex between this 15-mer L-dystroglycan peptide and a sub-domain of dystrophin has recently been determined [23] and shows that the L-dystroglycan peptide lies on the surface of dystrophin at a region formed by the WW domain and one of the `EF hands' of dystrophin (Fig. 6).
Fig. 5. Epitope mapping of MANHINGE4A mAb against dystrophin using a phage-displayed 15-mer peptide library. One selection experiment (a) identi¢ed ¢ve clones displaying di¡erent peptide sequences. Residues that match the original sequence (b) are shown in boldface and underlined.
BBAGEN 25192 13-6-01
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
Fig. 6. Position of the MANDAG and MANHINGE4A epitopes in the 3D structure of a dystrophin-L-dystroglycan complex. The structure of a C-terminal L-dystroglycan peptide complexed with a C-terminal dystrophin fragment ([24]; pdb accession 1EG4) was downloaded from http:// www.ncbi.nlm.nih.gov/entrez and rotated using the Cn3D 3.0 viewing program from the same website. Residue positions were identi¢ed using Cn3D and labelled in white using Adobe Photoshop. The dystroglycan peptide is at top right and the WW domain is the L-sheet below it.
The L-dystroglycan sequence is accessible on the surface of the complex and is essentially linear. The core residues of the MANHINGE4A epitope, PWxRA, form one strand of the three-stranded anti-parallel L-sheet that constitutes the WW domain (Fig. 6). This structure raises the interesting question of how antibody binding can be accommodated within it. When dystrophin and L-dystroglycan co-localize at the sarcolemma in muscle tissue sections, the PPxY motif and WW domain are interacting but the immunohistochemistry (Fig. 1) suggests that epitope recognition by the mAbs is not prevented by the interaction. The possibility that the mAbs are recognizing side-chains projecting from one accessible face is not supported by the 3D structure, which shows the side-chains of both PWxRA in dystrophin and PPPYV in dystroglycan projecting in different directions from the main chain. Linear epitopes have to become partially buried in the mAb for fully effective interaction of the mAb with their side-chains to occur. Accessibility is not a problem for Western blots, since the antigen will be disrupted and epitopes exposed by SDS treatment, but some disruption of the dystrophin^ dystroglycan structure could also required for mAb binding to occur in situ at the sarcolemma, as observed immunohistochemically. There are three possible hypotheses. One is that the mAbs do recognize the structure in Fig. 6, but only weakly since not all the side-chains of the epitopes are accessible; this is unconvincing, but cannot be ruled out. A second is that, during immunohistochemical procedures, L-dystroglycan mAbs could displace the dystrophin in situ and vice-versa. This might explain why the mAbs only recognize L-dystroglycan or dystrophin in un¢xed tissue sections (results not shown), since formalin cross-linking would prevent displacement. A third hypothesis is that the interaction in muscle is less than 100%, so that dystrophin molecules with free WW-domains and
59
L-dystroglycan molecules with free PPxY motifs are always available at the sarcolemma in vivo. This scenario has not been ruled out experimentally since dystrophin can be retained at the sarcolemma without the dystroglycan interaction; for example, by its N-terminal interactions with sub-sarcolemmal actin [25]. Finally, the four dystrophin `hinges' were originally identi¢ed as £exible regions by their susceptibility to proteolysis, both in vitro [26] and in situ [27]. Although the exact residues susceptible to proteolysis have not been identi¢ed, hinge 4 peptide sequences were chosen as likely immunogens because of their predicted £exibility [14]. It is a little surprising that the MANHINGE4 mAb obtained in this way turns out to recognize part of a L-sheet structure rather than the random coil sequence that precedes it. Acknowledgements This work was supported by grants from the Muscular Dystrophy Campaign and the Higher Education Funding Council (Wales) DevR scheme.
References [1] O. Ibraghimov-Beskrovnaya, J.M. Ervasti, C.J. Leveille, C.A. Slaughter, S.W. Sernett, K.P. Campbell, Primary structure of dystrophin-associated glycoproteins linking dystrophin to the extracellular matrix, Nature 355 (1992) 696^702. [2] J.F. Talts, Z. Andac, W. Gohring, A. Brancaccio, R. Timpl, Binding of the G domains of laminin alpha1 and alpha2 chains and perlecan to heparin, sulfatides, alpha-dystroglycan and several extracellular matrix proteins, EMBO J. 18 (1999) 863^870. [3] S. Rentschler, H. Linn, K. Deininger, M.T. Bedford, X. Espanel, M. Sudol, The WW domain of dystrophin requires EF-hands region to interact with beta-dystroglycan, J. Biol. Chem. 380 (1999) 431^442. [4] B. Yang, D. Jung, D. Motto, J. Meyer, G. Koretzky, K.P. Campbell, SH3 domain-mediated interaction of dystroglycan and Grb2, J. Biol. Chem. 270 (1995) 11711^11714. [5] E.P. Ho¡man, R.H. Brown Jr., L.M. Kunkel, Dystrophin: the protein product of the Duchenne muscular dystrophy locus, Cell 51 (1987) 919^928. [6] M. James, Nguyen thi Man, C.J. Wise, G.E. Jones, G.E. Morris, Utrophin-dystroglycan complex in membranes of adherent cultured cells, Cell Motil. Cytoskeleton 33 (1996) 163^174. [7] F. Sotgia, J.K. Lee, K. Das, M. Bedford, T.C. Petrucci, P. Macioce, M. Sargiacomo, F.D. Bricarelli, C. Minetti, M. Sudol, M.P. Lisanti, Caveolin-3 directly interacts with the C-terminal tail of beta-dystroglycan. Identi¢cation of a central WW-like domain within caveolin family members, J. Biol. Chem. 275 (2000) 38048^38058. [8] D. Jung, B. Yang, J. Meyer, J.S. Chamberlain, K.P. Campbell, Identi¢cation and characterization of the dystrophin anchoring site on beta-dystroglycan, J. Biol. Chem. 270 (1995) 27305^27310. [9] M. James, A. Nuttall, J.L. Ilsley, K. Ottersbach, J.M. Tinsley, M. Sudol, S.J. Winder, Adhesion-dependent tyrosine phosphorylation of (beta)-dystroglycan regulates its interaction with utrophin, J. Cell Sci. 113 (2000) 1717^1726. [10] J.M. Ervasti, K.P. Campbell, A role for the dystrophin-glycoprotein complex as a transmembrane linker between laminin and actin, J. Cell Biol. 122 (1993) 809^823.
BBAGEN 25192 13-6-01
60
A.V. Pereboev et al. / Biochimica et Biophysica Acta 1527 (2001) 54^60
[11] S.J. Winder, The membrane^cytoskeleton interface: the role of dystrophin and utrophin, J. Muscle Res. Cell Motil. 18 (1997) 617^629. [12] G.E. Morris, Epitope mapping, in: R. Rapley, J.M. Walker (Eds.), Molecular Biomethods Handbook, Humana Press, Totowa, NJ, 1998, pp. 619^630. [13] T.R. Helliwell, Nguyen thi Man, G.E. Morris, Expression of the 43 kD dystrophin-associated glycoprotein in human neuromuscular disease, Neuromusc. Disord. 4 (1994) 101^113. [14] N. Ahmed, Nguyen thi Man, G.E. Morris, Flexible hinges in dystrophin, Biochem. Soc. Trans 26 (1998) S310. [15] G.E. Morris, Nguyen thi Man, A. Pereboev, J. Kendrick-Jones, S.J. Winder, Disruption of the utrophin-actin interaction by monoclonal antibodies and prediction of an actin-binding surface of utrophin, Biochem. J. 337 (1999) 119^123. [16] Nguyen thi Man, G.E. Morris, A rapid method for generating large numbers of high-a¤nity monoclonal antibodies from a single mouse, in: J.M. Walker (Ed.), The Protein Protocols Handbook, Humana Press, Totowa, NJ, 1996, pp. 783^792. [17] Nguyen thi Man, G.E. Morris, Production of antibodies by the hybridoma method, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 377^389. [18] G.P. Smith, Cloning in fUSE vectors, Division of Biological Sciences, University of Missouri at Columbia, USA, 1992 (available directly from Prof. G.P. Smith). [19] A. Pereboev, G.E. Morris, Reiterative screening of phage display peptide libraries, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 195^206. [20] D. Lederfein, Z. Levy, N. Augier, J. Leger, G.E. Morris, O. Fuchs, D. Ya¡e, U. Nudel, 71 kD protein is a major product of the Du-
[21]
[22] [23]
[24]
[25]
[26]
[27]
[28]
chenne Muscular Dystrophy gene in brain and other non-muscle tissues, Proc. Natl. Acad. Sci. USA 89 (1992) 5346^5350. Nguyen thi Man, J.M. Ellis, I.B. Ginjaar, M.M.B. van Paassen, G.J.B. van Ommen, A.F.M. Moorman, A.J. Cartwright, G.E. Morris, Monoclonal antibody evidence for structural similarities between the central rod regions of actinin and dystrophin, FEBS Lett. 272 (1990) 109^112. van Regenmortel, Structural and functional approaches to the study of protein antigenicity, Immunol. Today 10 (1989) 266^272. F.A. Saul, P.M. Alzari, Crystallographic studies of antigen-antibody interactions, in: Epitope Mapping Protocols, Humana Press, Totowa, NJ, 1996, pp. 11^23. X. Huang, F. Poy, R. Zhang, A. Joachimiak, M. Sudol, M.J. Eck, Structure of a WW domain containing fragment of dystrophin in complex with beta-dystroglycan, Nat. Struct. Biol. 7 (2000) 634^638. T.R. Helliwell, J.M. Ellis, R.C. Mountford, R.E. Appleton, G.E. Morris, A truncated dystrophin lacking the C-terminal domains is localized at the muscle membrane, Am. J. Hum. Genet. 50 (1992) 508^514. M. Koenig, L.M. Kunkel, Detailed analysis of the repeat domain of dystrophin reveals four potential hinge segments that may confer £exibility, J. Biol. Chem. 265 (1990) 4560^4566. S. Hori, S. Ohtani, Nguyen thi Man, G.E. Morris, The N-terminal half of dystrophin is protected from proteolysis in situ, Biochem. Biophys. Res. Commun. 209 (1995) 1062^1067. G.E. Morris, S.G. Sedgwick, J.M. Ellis, A. Pereboev, J.S. Chamberlain, Nguyen thi Man, An epitope structure for the C-terminal domain of dystrophin and utrophin, Biochemistry 37 (1998) 11117^ 11127.
BBAGEN 25192 13-6-01