Nội dung

Concerted mutation of Phe residues belonging to the b-dystroglycan ectodomain strongly inhibits the interaction with a-dystroglycan in vitro Manuela Bozzi1,*, Francesca Sciandra2,*, Lorenzo Ferri2, Paola Torreri3, Ernesto Pavoni2, Tamara C. Petrucci3, Bruno Giardina2 and Andrea Brancaccio2 1 Istituto di Biochimica e Biochimica Clinica, Università Cattolica del Sacro Cuore, Rome, Italy 2 CNR, Istituto di Chimica del Riconoscimento Molecolare c ⁄ o Istituto di Biochimica e Biochimica Clinica, Università Cattolica del Sacro Cuore, Rome, Italy 3 Dipartimento di Biologia Cellulare e Neuroscienze, Istituto Superiore di Sanità, Rome, Italy Keywords alanine scanning; cell transfection; dystroglycan; protein–protein interaction; site-directed mutagenesis Correspondence A. Brancaccio, CNR, Istituto di Chimica del Riconoscimento Molecolare c ⁄ o Istituto di Biochimica e Biochimica Clinica, Università Cattolica del Sacro Cuore, Largo Francesco Vito 1, 00168 Rome, Italy Fax: +39 6 3053598 Tel: +39 6 3057612 E-mail: andrea.brancaccio@icrm.cnr.it *These authors contributed equally to this work (Received 10 July 2006, revised 1 September 2006, accepted 6 September 2006) doi:10.1111/j.1742-4658.2006.05492.x The dystroglycan adhesion complex consists of two noncovalently interacting proteins: a-dystroglycan, a peripheral extracellular subunit that is extensively glycosylated, and the transmembrane b-dystroglycan, whose cytosolic tail interacts with dystrophin, thus linking the F-actin cytoskeleton to the extracellular matrix. Dystroglycan is thought to play a crucial role in the stability of the plasmalemma, and forms strong contacts between the extracellular matrix and the cytoskeleton in a wide variety of tissues. Abnormal membrane targeting of dystroglycan subunits and ⁄ or their aberrant post-translational modification are often associated with several pathologic conditions, ranging from neuromuscular disorders to carcinomas. A putative functional hotspot of dystroglycan is represented by its intersubunit surface, which is contributed by two amino acid stretches: approximately 30 amino acids of b-dystroglycan (691–719), and approximately 15 amino acids of a-dystroglycan (550–565). Exploiting alanine scanning, we have produced a panel of site-directed mutants of our two consolidated recombinant peptides b-dystroglycan (654–750), corresponding to the ectodomain of b-dystroglycan, and a-dystroglycan (485–630), spanning the C-terminal domain of a-dystroglycan. By solidphase binding assays and surface plasmon resonance, we have determined the binding affinities of mutated peptides in comparison to those of wildtype a-dystroglycan and b-dystroglycan, and shown the crucial role of two b-dystroglycan phenylalanines, namely Phe692 and Phe718, for the a–b interaction. Substitution of the a-dystroglycan residues Trp551, Phe554 and Asn555 by Ala does not affect the interaction between dystroglycan subunits in vitro. As a preliminary analysis of the possible effects of the aforementioned mutations in vivo, detection through immunofluorescence and western blot of the two dystroglycan subunits was pursued in dystroglycan-transfected 293-Ebna cells. Dystroglycan (DG) is an adhesion molecule composed of two subunits, a-DG and b-DG [1], encoded by a single gene, dag1, which produces a unique polypeptide precursor consisting of 895 amino acids. A post-translational cleavage, performed by a still unidentified protease at the Gly653-Ser654 site, produces two subunits, Abbreviations DG, dystroglycan; DGC, dystroglycan–glycoprotein complex; EGFP, enhanced green fluorescent protein; SPR, surface plasmon resonance. FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS 4929 Mutagenesis at the a–b dystroglycan interface M. Bozzi et al. a-DG and b-DG. a-DG is a highly glycosylated peripheral membrane protein that interacts with several extracellular matrix proteins such as laminin, perlecan and agrin [2]. b-DG spans the membrane and binds a-DG in a noncovalent way, providing a connection between the extracellular matrix and the cytoskeleton inside the cells, where it interacts with dystrophin, utrophin and other cytosolic proteins, such as rapsyn, caveolin-3 and Grb2 [3–5]. Together with sarcoglycans, dystrobrevins, syntrophins, and sarcospan, DG forms the dystrophin–glycoprotein complex (DCG), which plays an essential role as a scaffold for cells in muscle and in a wide variety of nonmuscle tissues [6,7], including the central and peripheral nervous systems, and several epithelial tissues [8]. The importance of the DCG is dramatically apparent in several forms of muscular dystrophy, where mutations in DCG proteins lead to instability and progressive weakness of the muscle fibers [9]. Although no natural mutations have been detected in DG, it is substantially altered or absent in muscular dystrophies. For this reason, detailed molecular characterization of the subunit interface should be considered of primary importance for our understanding of the overall stability of the DGG, and perhaps in the future for surgical modulation of its function with the purpose of alleviating severe human diseases [10]. Primary structure analysis and electron microscopy have shown that a-DG has a dumbbell-like structure organized in two globular domains, the N-terminal and C-terminal domains, connected by an elongated central mucin-like region that contains highly glycosylated sequences, rich in prolines, serines and threonines [11]. A structural characterization of the N-terminal domain was recently obtained by a crystallographic analysis carried out on a murine a-DG N-terminal fragment, and revealed the presence of two autonomous modules connected by a long and flexible linker. The N-terminal module shows Ig-like folding, whereas the C-terminal module appears to be very similar to the ribosomal RNA-binding proteins [12]. The only structural hints concerning the C-terminal domain of a-DG come from a sequence alignment approach, which has shown some similarities with cadherin domains [13]. Previous studies, carried out employing a series of independent techniques such as IR, CD [14] and NMR spectroscopy [15], have revealed the absence of any classic secondary structural element in the recombinant b-DG ectodomain, which shows high conformational plasticity, typical of a natively unfolded protein. The noncovalent interaction between the two DG subunits occurs between the C-terminal region of a-DG and the N-terminal ectodomain of b-DG, and is 4930 apparently independent of glycosylation [16]. Solidphase binding assays, performed with recombinant fragments corresponding to the C-terminal domain of a-DG harboring progressive deletions, have shown that the b-DG-binding epitope resides between amino acids 550 and 585 [17], and further NMR analysis has narrowed this location to amino acids 550–565 [18]. In addition, extensive NMR structural characterization of our 15N ⁄ 13C b-DG(654–750) recombinant fragment, spanning the b-DG ectodomain, suggested that the a-DG-binding epitope corresponds to an amino acid stretch located between positions 691 and 719 [15]. In order to identify the specific amino acids within the linear interacting epitopes involved in the complex between a-DG and b-DG, alanine scanning of some of the residues that were mainly influenced in NMR titrations [15] was performed on recombinant fragments a-DG(485–630) and b-DG(654–750). The reciprocal affinities of wild-type a-DG and b-DG peptides vs. the panel of mutated recombinant fragments were measured using two independent techniques: solid-phase binding assays, exploiting biotinylated recombinant ligands, and surface plasmon resonance (SPR), in which one peptide is covalently immobilized on a sensor chip while the other is used in soluble phase without labels. Such mutations were also imported into full-length DG constructs cloned into an appropriate vector to transfect eukaryotic cells, in order to set up a suitable cellular system to study the effect of site-directed mutagenesis on the processing and targeting of DG in vivo. Results Mutations within the b-DG(654–750) recombinant fragment The phenylalanine in position 718, belonging to the putative a-DG-binding epitope (691–719) within the ectodomain of b-DG, was found to be one of the most influenced residues during the titration of [15N]b-DG(654–750) with thioredoxin-a-DG(485–620) [15]. Therefore, we decided to mutate it to an alanine, together with two other phenylalanines, Phe692 and Phe700, the only other aromatic residues located within the a-DG-binding epitope that are highly conserved in all the species so far analyzed. A more drastic alteration of the protein primary structure was produced by deleting six amino acids, located within the a-DG-binding epitope, between positions 701 and 706. A previous NMR characterization of the b-DG ectodomain [15] revealed that the amino acids between positions 701 and 704 are so flexible as to be undetectable under the experimental conditions used FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS M. Bozzi et al. for NMR analysis. We believed that it would be interesting to verify whether such a flexible amino acid stretch, located within the a-DG-binding epitope, might play a role in the interaction between the a-DG and b-DG subunits. Three additional mutations were introduced outside the putative a-DG-binding epitope to check whether perturbing the b-DG ectodomain elsewhere might also influence its interaction with a-DG. We produced two mutations upstream of the a-DG-binding epitope, such as Trp659 fi Ala, because Trp659 is the only aromatic residue in this portion of the protein, and Glu667 fi Ala; only one mutation, Val736 fi Ala, was generated within the C-terminal region of the b-DG ectodomain, downstream of the a-DG-binding epitope. A map of all the mutations produced within the recombinant fragments b-DG(654–750) and a-DG(485–620) is given in Fig. 1. In order to measure the affinity of such mutants for a-DG(485–630), a series of solid-phase binding assays was carried out. Typically, in solid-phase binding assays, a-DG(485–630) was coated onto microtiter plates, whereas b-DG(654–750) and its b-DG(654– mutants, b-DG(654–750)Trp659 fi Ala, Glu667 fi Ala , b-DG(654–750)Phe692 fi Ala, b-DG(654– 750) 750)Phe700 fi Ala, b-DG(654–750)Phe718 fi Ala, b-DG(654– 750)Val736 fi Ala, b-DG(654–750)Phe692 fi Ala ⁄ Phe718 fi Ala, b-DG(654–750)Phe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala and b-DG(654–750)D(701)706), were biotinylated and used as soluble ligand at increasing concentrations (up to 20 lm). Fig. 1. A panel of mutations hitting the reciprocal a-DG–b-DG binding epitopes was generated. In the C-terminal region of a-DG, between amino acids 550 and 565, Trp551, Phe554 and Asn555 were mutated to alanine. In the a-DG-binding epitope comprising residues 691–719 of the b-DG ectodomain, the mutations Phe692 fi Ala, Phe700 fi Ala and Phe718 fi Ala were generated while the residues from 701 to 706 were knocked-in. Three additional mutations were introduced: Glu667 fi Ala and Trp659 fi Ala upstream, and Val736 fi Ala downstream, of the a-DG-binding epitope. Mutagenesis at the a–b dystroglycan interface The apparent affinity for a-DG(485–630), exhibited by the mutants b-DG(654–750)Glu667 fi Ala and b-DG(654–750)Val736 fi Ala and evaluated by solid-phase binding assays, was very similar to that displayed by b-DG(654–750) (Fig. 2A, Table 1), whereas all the other single mutants, namely b-DG(654–750)Trp659 fi Ala, b-DG(654–750)Phe692 fi Ala, b-DG(654–750)Phe700 fi Ala and b-DG(654–750)Phe718 fi Ala, showed reduced affinity for a-DG(485–630) (Fig. 2B). Also, the deletion mutant b-DG(654–750)D(701)706) was able to bind a-DG(485– 630) with the same affinity as the wild type, demonstrating that the highly flexible stretch corresponding to positions 701–706 is not involved in the interaction with a-DG(485–630) and does not alter significantly the b-DG ectodomain conformation (Fig. 2A, Table 1). On the other hand, double and triple mutations, such as Phe692 fi Ala ⁄ Phe718 fi Ala and Phe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala, completely abolished the binding between b-DG(654–750) and a-DG(485– 630), at least in the ligand concentration range explored (Fig. 2C). To rule out the possibility that the lower affinity for a-DG(485–630) exhibited by some b-DG mutants might be due to some major proteolytic event, all the samples used to perform solid-phase binding assays were checked by Tricine ⁄ SDS ⁄ PAGE before and after biotinylation, and showed the same mobility as wild-type b-DG(654–750), indicating that no degradation occurred within mutated recombinant fragments; similarly, we did not observe any evident aggregation behavior when analyzing the various mutated b-DG peptides by native gel electrophoresis (data not shown). The solid-phase binding assay data were confirmed by SPR experiments, in which a-DG(485–630) was immobilized on a sensor chip, and b-DG(654–750) and its mutants were used in soluble phase as analytes. First, the dissociation equilibrium constant KD was measured for the interaction between the two wildtype recombinant fragments a-DG(485–630) and b-DG(654–750). It should be noted that the affinity constant value for the b-DG(654–750)–a-DG(485–630) interaction, measured by immobilizing b-DG(654–750) and using a-DG(485–630) as analyte (KD 2.73 lm) (Table 1, Supplementary Fig. S1), was fully comparable to the value obtained when a-DG(485–630) was immobilized and b-DG(654–750) was used as analyte (KD 2.66 lm) (Table 1). The thermodynamic constant KD was also measured for the interaction between the wild-type recombinant fragment a-DG(485–630) and the mutant b-DG(654–750)Phe700 fi Ala, and confirmed its reduced affinity for a-DG(485–630) with respect to the wild type (KD 7.00 lm; Table 1). The kinetic SPR profiles obtained for all the single mutants b-DG(654–750)Phe692 fi Ala, b-DG(654– FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS 4931 Mutagenesis at the a–b dystroglycan interface M. Bozzi et al. Table 1. (A) Equilibrium dissociation constants (KD) calculated by solid-phase binding assays and SPR. Mean apparent KD values and relative standard deviations, calculated for the interaction between wild-type and mutated recombinant fragments, b-DG(654–750) and a-DG(485–630), by solid-phase binding assays. The values are averaged over a number of independent experiments, indicated in parentheses. For the b-DG mutants showing reduced affinity for a-DG(485–630), KD values cannot be calculated (ND, not determined; see Experimental procedures). (B) KD values for the interaction between wild-type recombinant fragments, b-DG(654–750) and a-DG(485–630), and b-DG(654–750)Phe700 fi Ala and a-DG(485–630), as measured by SPR. (A) Solid-phase binding assays A Immobilized protein ⁄ biotinylated protein wt a-DG ⁄ b-DG a-DGwt ⁄ b-DGGlu667 fi Ala a-DGwt ⁄ b-DGD(701–706) a-DGwt ⁄ b-DGVal736 fi Ala a-DGwt ⁄ b-DGTrp659 fi Ala a-DGwt ⁄ b-DGPhe692 fi Ala a-DGwt ⁄ b-DGPhe700 fi Ala a-DGwt ⁄ b-DGPhe718 fi Ala a-DGwt ⁄ b-DGPhe692 fi Ala ⁄ Phe718 fi Ala a-DGwt ⁄ b-DGPhe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala a-DGTrp551 fi Ala ⁄ b-DGwt a-DGPhe554–Ala ⁄ b-DGwt a-DGAsn555 fi Ala ⁄ b-DGwt a-DGTrp551 fi Ala ⁄ Phe554 fi Ala ⁄ b-DGwt B Apparent KD (lM) 2.8 ± 0.9 (9) 3.5 ± 0.9 (3) 2.9 ± 0.8 (3) 2.9 ± 2 (4) ND (10) ND (8) ND (3) ND (8) ND (3) ND (3) 1.3 ± 0.8 (4) 2.3 ± 0.9 (5) 1.5 ± 0.8 (6) 2.4 ± 0.7 (3) (B) SPR C Fig. 2. Solid-phase binding assays. a-DG(485–630) was immobilized on plates, whereas b-DG(654–750) (black) and its mutants b-DG(654–750)Glu667 fi Ala (blue), b-DG(654–750)Val736 fi Ala (red), b-DG(654–750)D(701)706) (green) (A), b-DG(654–750)Trp659 fi Ala (magenta), b-DG(654–750)Phe692 fi Ala (green), b-DG(654– Phe700 fi Ala 750) (blue), b-DG(654–750)Phe718 fi Ala (yellow) (B), b-DG(654–750)Phe692 fi Ala ⁄ Phe718 fi Ala (cyan), and b-DG(654– 750)Phe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala (red) (C), were used as biotinylated ligands. Every point is an average of three or more independent experiments. The continuous line represents fitting of experimental data using a single class of equivalent binding sites equation. The maximal binding of control b-DG(654–750), extrapolated by fitting experimental data, was set as 100% (see Experimental procedures). 4932 wt Immobilized protein ⁄ analyte KD (lM) a-DGwt ⁄ b-DGwt b-DGwt ⁄ a-DGwt a-DGwt ⁄ b-DGPhe700 fi Ala 2.66 2.73 7.00 750)Phe700 fi Ala and b-DG(654–750)Phe718 fi Ala were in line with the reduction of affinity measured by solidphase binding assays, indicating a lower association rate and a higher dissociation rate with respect to wild-type a-DG. The only exception was the mutant b-DG(654–750)Phe700 fi Ala, which showed a higher association rate but also a higher dissociation rate in comparison to the wild type; the double mutant b-DG(654–750)Phe692 fi Ala ⁄ Phe718 fi Ala did not bind a-DG(485–630) at all (Fig. 3). Mutations within the a-DG(485–630) recombinant fragment Our previous NMR analysis using a synthetic peptide corresponding to a-DG(550–585) and the recombinant fragment b-DG(654–750) indicated that the residues between positions 550 and 565 of a-DG belong to an amino acid stretch that is likely to be involved in the FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS M. Bozzi et al. Mutagenesis at the a–b dystroglycan interface 630), indicating that the effect measured can be ascribed to the mutations within the b-DG ectodomain (data not shown). Interestingly, a western blot experiment showed that Trp551, Phe554 and Asn555 are also not likely to be key residues for the interaction with the mAb sx ⁄ 3 ⁄ 50 ⁄ 25 directed against the b-DG-binding epitope (residues 549–567 of a-DG) [19], as the antibody is able to recognize the bands relative to a-DG recombinant peptides in western blot experiments (Fig. 4A). Moreover, the antibody is also able to bind a-DG mutated peptides in solid-phase binding assays, as it inhibits the interaction between b-DG(654–750) and the mutant a-DG(485– 630)Phe554 fi Ala (Fig. 4B), as previously shown for wild-type a-DG(485–600) [19]. Fig. 3. SPR kinetic profiles of the interaction between immobilized a-DG(485–630) and b-DG(654–750) (black) and its mutants, b-DG(654–750)Phe692 fi Ala (green), b-DG(654–750)Phe700 fi Ala (blue), b-DG(654–750)Phe718 fi Ala (yellow), b-DG(654–750)Phe692–Ala ⁄ Phe718 fi Ala (cyan), used as analytes at a fixed concentration of 10 lM. interaction with b-DG(654–750), based on data collected at the level of their NH and CHa [18]. In order to test the role of individual amino acid side chains at the a-DG–b-DG interface, alanine scanning of positions Trp551, Phe554 and Asn555 was carried out. Solid-phase binding assays were carried out, in which a-DG(485–630) and its mutants, a-DG(485–630)Trp551 fi Ala, a-DG(485–630)Phe554 fi Ala, and a-DG(485– a-DG(485–630)Asn555 fi Ala 630)Trp551 fi Ala ⁄ Phe554 fi Ala were coated onto the microtiter plate and biotinylated b-DG(654–750) was used as soluble ligand. All the mutants showed the same affinity for b-DG(654–750) as the wild type, indicating that such mutations are not likely to have major effects on the interaction between a-DG and b-DG (Supplementary Fig. S2). Accordingly, the apparent dissociation constant values obtained by fitting the experimental data are very similar to the values calculated for the interaction between the wild-type peptides a-DG(485–630) and b-DG(654–750) (Table 1). To further confirm that residues Trp551, Phe554 and Asn555 are not involved in the interaction with b-DG, the affinities of some b-DG mutants for the mutants a-DG(485–630)Trp551 fi Ala, a-DG(485–630)Phe554 fi Ala and a-DG(485–630)Asn555 fi Ala were also estimated by solid-phase binding assays. The affinities of b-DG(654–750)Trp659 fi Ala, b-DG(654–750)Phe692 fi Ala, b-DG(654–750)Phe700 fi Ala, b-DG(654–750)Phe718 fi Ala and b-DG(654–750)Phe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala for immobilized a-DG mutants were very similar to the reduced affinity exhibited for wild-type a-DG(485– A B Fig. 4. (A) Western blot on 12% SDS ⁄ PAGE of wild-type and mutated recombinant fragments of a-DG. Recombinant fragments were detected using mAb sx ⁄ 3 ⁄ 50 ⁄ 25. Lane 1: a-DG(485– 630)Trp551–Ala ⁄ Phe554 fi Ala. Lane 2: a-DG(485–630)Asn555 fi Ala. Lane 3: a-DG(485–630)Phe554 fi Ala. Lane 4: a-DG(485–630)Trp551 fi Ala. Lane 5 (control): a-DG(485–630). (B) Solid-phase binding assays were performed by immobilizing a-DG(485–630) (d) and a-DG(485– 630)Phe554 fi Ala (h) and using biotinylated b-DG(654–750) as soluble ligand, in the presence (empty symbols) and in the absence (full symbols) of mAb sx ⁄ 3 ⁄ 50 ⁄ 25. FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS 4933 Mutagenesis at the a–b dystroglycan interface M. Bozzi et al. Transfection of 293-Ebna cells with wild-type and mutated DG constructs In order to verify the correct membrane targeting of mutated DG, DNA constructs spanning the entire DG gene, including its signal peptide, and carrying the mutations analysed in vitro, such as Trp551 fi Ala, Phe554 fi Ala, Asn555 fi Ala, Glu667 fi Ala, Phe692 fi Ala, Phe700 fi Ala, Phe718 fi Ala, Phe692 fi Ala ⁄ Phe718 fi Ala and Val736 fi Ala, were included in an appropriate mammalian expression vector and then transfected into human 293-Ebna cells. The cytomegalovirus promoter drives the efficient transcription of the DG exogenous gene, which was strongly expressed in the transfected cells (Fig. 5). None of the mutations seemed to significantly alter the A Fig. 5. Immunostaining of wild-type DG and its mutants in transiently transfected 293-Ebna cells. (A) The a-subunits were stained with a polyclonal antibody directed against the C-terminal domain of a-DG on intact 293-Ebna cells. (B) Detection of b-DG was carried out using b-DG antibody in permeabilized 293-Ebna cells. Both subunits of wild-type and mutated DG were clearly overexpressed with respect to nontransfected cells displaying a much lower and diffuse staining due to endogenous DG; the double mutation Phe692 fi Ala ⁄ Phe718 fi Ala does not alter the membrane targeting of a-DG. 4934 FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS M. Bozzi et al. Mutagenesis at the a–b dystroglycan interface B Fig. 5. (Continued). correct membrane localization of a-DG and b-DG, at least 24 h after transfection, as all the mutants could be stained with a polyclonal antibody directed against the C-terminal region of a-DG [20] and with a commercial antibody directed against the cytoplasmic tail of b-DG (Fig. 5A,B). Also, the double mutation Phe692 fi Ala ⁄ Phe718 fi Ala, which greatly reduces the affinity of b-DG for the a-subunit in vitro, did not influence the localization of the two DG subunits (Fig. 5A,B). In order to detect any effect of the double mutation Phe692 fi Ala ⁄ Phe718 fi Ala, which may have evaded the immunostaining analysis [21], the entire DG carrying this mutation was cloned into the pEGFP vector, which codes for the enhanced green fluorescence protein (EGFP) fused at the C-terminus of b-DG. This vector was used to transiently transfect 293-Ebna cells. Other two pEGFP vectors were produced, carrying wild-type DG and its mutant DGPhe554 fi Ala, respectively, to be used as a control (Fig. 6). EGFP increases the molecular mass of b-DG by 25 kDa, allowing us to distinguish between the FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS 4935 Mutagenesis at the a–b dystroglycan interface M. Bozzi et al. A B C D exogenous b-DG and the endogenous b-DG in western blot experiments. Western blot analysis carried out on total protein extracts from 293-Ebna cells transiently transfected with wild-type and mutated (Phe554 fi Ala or Phe692 fi Ala ⁄ Phe718 fi Ala) DG genes did not show any aberrant processing or glycosylation patterns of DG. Although lower expression of a-DG was detected in all transfected cells (including those transfected with empty pEGFP or wild-type pDG– EGFP; Fig. 7A), the amount of a-DG in the cells transfected with the double-mutated (Phe692 fi Ala ⁄ Phe718 fi Ala) DG construct was similar to that measured in wild-type DG-transfected cells (Fig. 7). Discussion In vitro inhibition of the a-DG–b-DG interaction via Phe to Ala mutations within the ectodomain of b-DG We investigated the interaction between a-DG and b-DG recombinant peptides carrying a series of sitedirected mutations. We performed amino acid substitutions using alanine, because this residue does not show any propensity for a specific secondary structure, and therefore does not perturb the overall protein conformation while highlighting the role of the side chain functional group that it replaces [22,23]. This charac4936 Fig. 6. 293-Ebna cells transiently transfected with the vector pEGFP, empty (A) or containing DNA constructs corresponding to wild-type DG (B) and its mutants DGPhe554 fi Ala (C) and DGPhe692 fi Ala ⁄ Phe718 fi Ala (D), where enhanced green fluorescent protein (EGFP) was fused at the C-terminus of b-DG. All the images are magnified 10·. EGFP alone was uniformly distributed throughout the cytoplasm, whereas DG–EGFP and its two mutants were mostly localized around the cellular periphery. In order to better visualize the DG complex location at the cellular periphery, a 40· magnified image was obtained referring to wild-type DG–EGFP, which clearly shows the membrane targeting of the chimeric wild-type DG–EGFP construct [inset in (B), red arrowheads]. teristic makes alanine the amino acid of choice for site-directed mutagenesis [22–26]. For the first time, we have measured the affinity between recombinant peptides spanning the C-terminal domain of a-DG and the b-DG ectodomain by solid-phase binding assays and SPR, demonstrating that despite the intrinsic pitfalls of solid-phase binding techniques, extensively reviewed by Tangemann & Engel [27], the apparent KD values measured with our ‘two-step’ biotin enzymelinked streptavidin approach are in full agreement with those measured with an independent and accurate technique such as SPR, whose reliability has been demonstrated in recent years for protein–protein interactions displaying very low affinity [28]. Our SPR measurements simply demonstrate that the apparent constants that we have estimated with the solid-phase binding approach are reliable. In this case, it seems that even a complex displaying a relatively low affinity (KD about 3 lm; Table 1) can be populated, when rapidly washed without major perturbations and then titrated with streptavidin [17,29]. In previous studies, we had already estimated that the a-DG-binding epitope of b-DG might involve a relatively extended region of approximately 30 amino acids, between positions 691 and 719 [15]. In order to obtain a deeper insight into the a–b interface, we have introduced a series of mutations within the b-DG ectodomain located in three different protein regions: FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS M. Bozzi et al. A B Fig. 7. Immunoblot of total protein extracts from 293-Ebna cells nontransfected (lane 1) or transfected with the empty pEGFP vector (lane 2), or containing DG–EGFP (lane 3) and its mutants DGPhe554 fi Ala–EGFP (lane 4) and DGPhe692 fi Ala ⁄ Phe718 fi Ala–EGFP (lane 5). (A) a-DG bound to the plasma membrane was probed with the commercial antibody VIA4-1 (Upstate, Charlottesville, VA, USA). (B) The amount of b-DG–EGFP (68 kDa) compared with that of endogenous b-DG (43 kDa) was detected with the commercial antibody directed against the cytoplasmatic tail of b-DG (upper panel) and with antisera against EGFP (middle panel). Anti-actin serum was used as loading control (lower panel). upstream, downstream and inside the putative a-DGbinding epitope. By measuring the affinity of these mutants for the recombinant fragment a-DG(485–630), corresponding to the C-terminal domain of a-DG, via solid-phase binding assays and SPR experiments, we have found three different behaviors: some mutants, b-DG(654– namely b-DG(654–750)Glu667 fi Ala, Val736 fi Ala and b-DG(654–750)D(701)706), show the 750) same affinity for a-DG(485–630) as the wild-type recombinant fragment; a second intermediate group, comprising b-DG(654–750)Trp659 fi Ala, b-DG(654– b-DG(654–750)Phe700 fi Ala and 750)Phe692 fi Ala, Phe718 fi Ala , shows a reduced affinity b-DG(654–750) for a-DG(485–630); the double and triple mutants, such as b-DG(654–750)Phe692 fi Ala ⁄ Phe718 fi Ala and b-DG(654–750)Phe692 fi Ala ⁄ Phe700 fi Ala ⁄ Phe718 fi Ala are completely unable to bind a-DG(485–630). The behavior of b-DG(654–750)Glu667 fi Ala and b-DG(654–750)Val736 fi Ala is not surprising, as these point mutations are located upstream and downstream Mutagenesis at the a–b dystroglycan interface of the a-DG-binding epitope, respectively, in portions of the protein that are not involved in the formation of the a–b interface [15]. However, this simple argument cannot be applied to explain the reduced affinity for a-DG(485–630) shown by the mutant b-DG(654–750)Trp659 fi Ala, whose amino acid substitution is located upstream of the a-DG-binding epitope. A possible interpretation of this result can be suggested on the basis of previous studies showing that the recombinant fragment b-DG(654–750) is organized into an N-terminal region, consisting of approximately 70 amino acids, which is characterized by restricted conformational mobility, and a highly flexible C-terminal region of approximately 30 residues [15]. In this context, an amino acid substitution introduced within the region of restricted mobility, such as Trp659 fi Ala, although located outside the putative a-DG-binding epitope, may perturb the b-DG(654– 750) conformational equilibrium, driving it to adopt non-native conformations unable to efficiently bind a-DG(485–630). All the point mutations located inside the putative a-DG-binding epitope, namely b-DG(654– b-DG(654–750)Phe700 fi Ala and 750)Phe692 fi Ala, Phe718 fi Ala , result in reduced affinity b-DG(654–750) between a-DG and b-DG recombinant peptides, which completely lose their ability to interact when two phenylalanines, Phe692 and Phe718, are simultaneously substituted with alanine. Surprisingly, the deletion of six residues corresponding to amino acids 701–706, located between the two important phenylalanines Phe692 and Phe718, does not perturb the interaction between a-DG and b-DG recombinant peptides, as can be deduced by comparing the values of the apparent KD for the b-DG(654–750)wt–a-DG(485–630) interaction and the b-DG(654–750)D(701)706)–aDG(485–630) interaction, as measured by solid-phase binding assays (Table 1). It should be noted that the choice to delete these specific amino acids (701–706) is based on the observation that the residues between positions 701 and 704, although belonging to the region of restricted mobility, are so flexible to be undetectable under the experimental conditions used for NMR experiments [15]. The stretch 701–706 may be part of a flexible linker that could bring two separate regions of the a-DGbinding epitope (carrying Phe692 and Phe718, respectively) closer to each other when they bind a-DG. Apparently, deleting the amino acid stretch 701–706 has no effect on the a–b interaction, so it is likely that the knock-in does not significantly alter the spatial distance between the two important phenylalanines. FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS 4937 Mutagenesis at the a–b dystroglycan interface M. Bozzi et al. Our previous results indicate that although the ectodomain of b-DG is a natively unfolded protein, its conformation is still maintained by a delicate network of long-range reciprocal interactions that govern major structural–functional events [30], to which the two phenylalanine residues, which are quite distant within the b-DG ectodomain linear sequence, may make an important contribution. When the binding experiments were performed with a-DG(485–630) and its mutants, we found that Trp551, Phe554 and Asn555 are not key residues for the interaction with b-DG(654–750), because no reduction of the affinity of the mutants a-DG a-DG(485–630)Phe554 fi Ala, (485–630)Trp551 fi Ala, Asn555 fi Ala and a-DG(485– a-DG(485–630) 630)Trp551 fi Ala ⁄ Phe554 fi Ala for b-DG(654–750), compared to wild-type a-DG(485–630), was measured by solid-phase binding assays (Supplementary Fig. S2). It should be pointed out that our previous NMR experiments showed that a-DG residues Trp551, Phe554 and Asn555, among others, were significantly influenced by the presence of b-DG(654–750) at the level of the protein backbone (i.e. at their NH and CHa), although at that time no data were collected on their side chains, which therefore could be substantially unaffected by b-DG binding [18], as is now strongly suggested by the results herein presented. A possible implication of our results is that the specificity of a-DG(485–630) binding to b-DG(654– 750) could depend mainly on its local conformation and only to a lesser extent on the chemical nature of the amino acid side chains involved in the binding. To further analyze this hypothesis, it will be necessary to introduce, within residues 550–565 of a-DG, amino acids that require stringent steric constraints, such as proline or isoleucine, which may significantly perturb the local structural characteristics of the b-DG-binding epitope. Interestingly, the amino acid substitutions that we analyzed do not impair binding to a monoclonal antibody, mAb sx ⁄ 3 ⁄ 50 ⁄ 25, as suggested by the western blot in Fig. 4A. Moreover, mAb sx ⁄ 3 ⁄ 50 ⁄ 25 is able to efficiently inhibit the interaction between wild-type a-DG and b-DG peptides [19] and also between mutated a-DG and b-DG (Fig. 4B), suggesting that other a-DG residues, belonging to the approximately 550–565 residues epitope, but located in the C-terminal region downstream of those so far analyzed, may be involved in the interaction both with mAb sx ⁄ 3 ⁄ 50 ⁄ 25 and b-DG(654–750). Therefore, additional amino acid substitutions, such as Ser556 fi Ala, Gly563 fi Ala and Pro565 fi Ala, will be tested in the future [18]. 4938 Analysis of DG subunit targeting in vivo: detection of DG subunits in 293-Ebna cells Few studies are currently available that focus on the identification and analysis of polymorphisms and ⁄ or point mutations within the DG gene in human populations [31,32]; nevertheless, it has been shown in a cellular system that single mutations may strongly influence the processing of the DG precursor and targeting at the cell surface of its subunits [33]. Mutations affecting some putative glycosylation sites lead to impaired membrane targeting of DG and also to defective cleavage of the precursor into the two subunits [33]. Furthermore, mutation at the precursor maturation cleavage site (Gly653-Ser654) induces muscular dystrophy in mice [34]. To investigate whether the amino acid substitutions tested in vitro influence the stability and the localization of DG at the plasma membrane, we transfected 293-Ebna cells with the full-length murine DG gene harboring the mutations Trp551 fi Ala, Phe554 fi Ala, Phe692 fi Ala, Asn555 fi Ala, Glu667 fi Ala, Phe700 fi Ala, Phe718 fi Ala, Val736 fi Ala and Phe692 fi Ala ⁄ Phe718 fi Ala. Cell-staining experiments showed that none of these mutations significantly affects the subcellular trafficking and plasmalemmal targeting of exogenous murine DG in transfected human 293-Ebna cells (Fig. 5). In fact, all the mutants were overexpressed, and a strong fluorescent signal was detected at the plasmalemma, exploiting a monoclonal antibody directed against the cytoplasmic tail of b-DG. Apparently, even the double mutation Phe692 fi Ala ⁄ Phe718 fi Ala, which greatly impairs binding between the two DG subunits in vitro, has no evident effect on the plasmalemmal targeting of b-DG (Fig. 5B). The correct membrane targeting of DG was also confirmed by the immunodetection of the a-subunit in human cells transfected with all the mutants (Fig. 5A). In order to further analyze possible effects of the double mutation Phe692 fi Ala ⁄ Phe718 fi Ala in vivo, we also cloned the mutant DGPhe692 fi Ala ⁄ Phe718 fi Ala within the pEGFP vector, together with wild-type DG and the mutant DGPhe554 fi Ala as controls (Fig. 6). Western blot analysis of total protein extracts from 293-Ebna cells, transiently transfected with these constructs, confirmed the results of cell-staining experiments (Fig. 7). The amount of a-DG in cells transfected with the mutant DGPhe692 fi Ala ⁄ Phe718 fi Ala is in fact fully comparable with that observed in cells transfected with the empty pEGFP vector or containing wild-type DG or DGPhe554 fi Ala (Fig. 7A). The reduced amount of a-DG detected by VIA4-1 in all the 293-Ebna transfected cell lines probably results from reduction and ⁄ or modifica- FEBS Journal 273 (2006) 4929–4943 ª 2006 The Authors Journal compilation ª 2006 FEBS