Portal de Periódicos da CAPES

Insight into interactions of the von-Willebrand-factor-A-like domain 2 with the FNIII-like domain 9 of collagen VII by NMR and SPR

2011; Wiley; Volume: 585; Issue: 12 Linguagem: Inglês

10.1016/j.febslet.2011.04.071

1873-3468

Sarah Leineweber, Sarah Schönig, Karsten Seeger,

Autoimmune Bullous Skin Diseases

mvWFA2 binds to mFNIII-9 by surface plasmon resonance (View interaction) Type VII collagen (Col7) is the main component of the anchoring fibrils, connecting the basal lamina with the tissue beneath by interacting with other collagen types and parts of the extracellular matrix [1, 2]. Col7 is a homotrimer consisting of one central collagenous domain interrupted by a short non-collagenous region. The central domain is flanked by two non-collagenous domains. The C-terminal non-collagenous domain 2 (NC-2) contains conserved cysteine residues which accompany an antiparallel assembly of these collagen molecules. In contrast the N-terminal non-collagenous domain 1 (NC-1) consists of several subdomains with high homologies to adhesion proteins. The C-terminal subdomain of NC-1 reveals high homology to von-Willebrand-factor A and is called von-Willebrand-factor-A like domain 2 (vWFA2) [3]. A schematic representation of a single Col7 α-chain forming a homotrimer is shown in Fig. 1 . Investigations have shown that vWFA2 plays a crucial role for binding to type I collagen but also for interactions with other components of the extracellular matrix [2, 4]. It is assumed that these connections contribute to skin stability. However, detailed knowledge of the skin architecture and the role of Col7 are still sparse and no high resolution structural information at atomic level is available. Col7 is involved in different skin blistering diseases. A heritable form is caused by mutations in type VII collagen, which are also located within vWFA2 [5]. Loss of tolerances to Col7 leads to the autoimmune skin blistering disease epidermolysis bullosa acquisita (EBA) characterized by autoantibodies against Col7. During the course of disease the dermis separates from the epidermis resulting in blisters and lesions [6]. Former investigations have shown that antigenic epitopes can be found also in vWFA2 [7, 8]. So far only little is known about the exact contribution of this domain in the architecture of the extracellular matrix and its role in developing EBA. Therefore, structural investigations of the murine von-Willebrand-factor-A-like domain 2 (mvWFA2) should help to gain more insight into the connection of the skin layers and the role of this domain in skin blistering diseases. Nuclear magnetic resonance (NMR) spectroscopy is advantageous for structural investigations due to its possibility of investigating dynamic systems as interactions with possible ligands. Based on NMR resonance assignment secondary structure elements have been obtained with the program torsion angle likelihood obtained from shift and sequence similarity+ (TALOS)+ [9]. Additional information has been achieved by 3 J-H(N)–H(α)-coupling constants and hydrogen–deuterium exchange. The experimental data are in good agreement with a homology model. Interestingly, an interaction with its neighboring domain the murine fibronectin-III-like domain 9 (mFNIII-9) has been shown the first time and was characterized by surface plasmon resonance (SPR) and NMR spectroscopy. A codon optimized sequence of mvWFA2 and of mFNIII-9 of Col7 was commercially synthesized (Mr. Gene, Regensburg, Germany) and cloned in pTWIN1 (NEB, Frankfurt, Germany). Cloning of the genes differs in the position of their intein affinity-tags. Whereas mvWFA2 has an N-terminal affinity-tag the affinity-tag of mFNIII-9 is C-terminal. Protein expression in Escherichia coli ER2566 and purification of both proteins followed the IMPACT™-TWIN protocol (NEB). mvWFA2 has been labeled uniformly with 15N and 13C according to Marley et al. [10]. Selective labeling for the amino acids A, R, and V was done according to Griffey et al. [11]. Cloning, expression and purification of mFNIII-9 was performed analog to mvWFA2. NMR experiments have been performed on a Bruker Avance DRX 500 spectrometer equipped with a TCI cryoprobe at 298 K in 10 mM sodium phosphate buffer, pH 7.4 and 10% D2O. 3-(trimethylsilyl)propionic acid-d 4 sodium salt (TSP-d 4) was added as internal chemical shift reference. 13C and 15N chemical shifts have been referenced using the absolute frequency ratios [12]. A set of three dimensional triple resonance experiments namely CBCACONH, HNCA, HCCCONH, HBHACONH, HNCOCA, HNCO, HNHA and NOESY-HSQC have been performed. 1H,15N-HSQC spectra of 15N–13C mvWFA2 in 100% D2O have been recorded additionally to identify slow exchanging amide protons. 1H,15N-HSQC spectra have been recorded immediately after 1H–2H exchange and 3 days later. Spectra processing was done with Bruker Topspin 2.0. For data analysis and resonance assignment Sparky has been used. Scalar coupling constants have been calculated for information of flexibility of mvWFA2 from an HNHA experiment according to Vuister and Bax [13]. Secondary structure determination has been performed with TALOS+ on the eNMR web portal [9, 14]. The interaction site of mvWFA2 with mFNIII-9 was determined via chemical shift mapping as described by Farmer et al. with b and f representing bound and free state [15]. The molar ratio of mvWFA2 to mFNIII-9 was approximately 1:2 with a concentration of 100 μM labeled mvWFA2. A three dimensional homology model has been obtained from Swiss Model [16]. The homology model was built on the basis of an alignment of the amino-acid sequence of mvWFA2 with a similar protein of known structure. This template protein is the crystal structure of the von-Willebrand-factor-A3 domain in complex with a Fab fragment of IgG RU5 (PDB: 1FE8C) and the overall sequence identity is 20.1%. A BIAcore 3000 has been used for SPR measurements. mvWFA2 was dissolved in 10 mM sodium acetate buffer pH 4 and immobilized on a CM5 sensor chip resulting in 6000 resonance units. This would correspond to a theoretical RUmax of 1156 for mFNIII-9. Binding experiments were performed in triplicate at room temperature in 10 mM sodium phosphate buffer pH 7.4. Dilutions of mFNIII-9 ranged from 100 μM to 2 mM. The K D was determined via steady-state curve fitting analysis of BIAevaluation software assuming an 1:1 binding stoichiometry. Expression and purification of 15N and 15N, 13C-labeled mvWFA2 yielded a single band in the SDS–PAGE (Supplementary data) with a typical overall yield of 19 mg protein per liter minimal medium. mvWFA2 shows a well-resolved 1H,15N-HSQC spectrum indicating that the recombinant mvWFA2 is properly folded (Fig. 2 ). The sample was stable over several months as identical spectrum patterns attest. Backbone assignments were obtained using standard triple resonance experiments. The assignment of the backbone chemical shifts is 96% complete. The N-terminal residues belonging to the affinity-tag could not be assigned due to flexibility. Solely from the chemical shifts a first structural character a disulfide bridge was identified. SDS–gel electrophoresis under non-reducing conditions revealed those to be intramolecular (data not shown). These chemical shift assignments present the initial point for structure determination of mvWFA2 and for protein–protein interaction studies. Based on chemical shift assignments a secondary structure prediction of mvWFA2 employing TALOS+ via the eNMR web-portal was performed [9, 14]. This program predicts torsion angles from Cα, Cβ, N, H and Hα NMR backbone chemical shifts (see Supplementary data). The resulting prediction for mvWFA2 is consistent with idealized secondary structure elements for the majority of the residues. Only a short region in the middle part (residues 78–87) is less well defined making a clear conclusion difficult. Overall a β-strand is predicted for this region. A comparison of the derived prediction with the homologous von-Willebrand-factor-A3 domain (PDB: 1AO3) matches well. It shows a similar distribution of α-helical and β-sheet sequences but they differ in length of secondary structure elements (Fig. 3 c). With the help of an HNHA spectrum scalar coupling constants have been determined. These permit also to draw conclusions regarding the secondary structure. Small 3 J (NH,Hα) couplings below 6 Hz are typical for α-helical regions while J couplings larger than 8 Hz are characteristic for β-sheets. The calculated coupling constants are altogether in good agreement with the secondary structure prediction of TALOS+ as shown in Fig. 2a. Merely, a part in the middle of the protein (residue 78–87) disagrees with the structure prediction suggesting an α-helical region instead of a β-strand according to TALOS+. Also the TALOS+ prediction for this part was not well defined, thus this protein region may be in a flexible loop. To identify slow exchanging amide protons a 1H,15N-HSQC was recorded immediately after 1H–2H exchange and 3 days later. Both spectra showed few peaks and the number of peaks vary only slightly. We conclude that most amide protons are solvent accessible whereas amide protons showing a slow exchange are part of the inner, hydrophobic core of the protein. At this stage the homology model of mvWFA2 was used to interpret the data. According to the homology model these slow exchanging residues in fact reside in the core of the protein and are part of the β-sheets as illustrated in Fig. 4 a. The experimental results are consistent with our model. The non-collagenous domain 1 mediates the interaction with other skin components such as type I collagen or laminin 322. For a deeper understanding of these interactions knowledge of the quaternary structure of Col7 is required. To obtain a more detailed insight of the structure of Col7 investigations of interactions of mvWFA with other Col7 domains have to be performed. Thus the interaction between mvWFA2 and its neighboring domain mFNIII-9 was analyzed. A 1H,15N-HSQC spectrum of 15N-labeled mvWFA2 with an excess of unlabeled mFNIII-9 shows a shift for certain residues. A chemical shift mapping identified those residues with significant alteration of the chemical shift. We conclude that these residues interact with the mFNIII-9 domain. The spectra indicate fast exchange at the chemical shift timescale which points to a weak binding affinity of mFNIII-9 to mvWFA2. In order to quantitate the interaction of mFNIII-9 and mvWFA2 we employed SPR. The determined K D is 0.8 ± 0.3 mM and in agreement with the NMR experiments. The affected residues have been mapped onto the homology model of mvWFA2 as shown in Fig. 4b. The majority of the residues are located at the same area of the model thus identifying the interaction site of mvWFA2 with mFNIII-9. Residues showing small chemical shift variations at sites far away from the interaction site are considered to be not part of the binding surface. The binding site for mFNIII-9 comprises the first and the last helix of mvWFA2 that is also in concordance with the homology model. This finding is further supported in a biological context because in the native Col7 molecule the FNIII-9 domain resides at this site. However, the most affected residues in the homology model point inwards of the protein indicating a slight different conformation of the helices in vivo. Comparison of our model with other homologous collagen binding proteins shows that the putative binding sites for type I collagen as well as the cell adhesion mediating RGD-motif are opposite to the mFNIII-9 interaction site. This suggests that mFNIII-9 is not involved in interactions with type I collagen which is supported by Chen et al. who showed that the FNIII repeats 7–9 participate in binding laminin 322 [17]. Successful expression, purification and NMR resonance assignment of the von-Willebrand-factor-A-like domain 2 of murine type VII collagen is the basis for detailed structural information on Col7. For the first time a specific interaction between mvWFA2 and mFNIII-9 was shown and the binding site for mFNIII-9 on mvWFA2 was identified. The binding affinity is weak but in the native molecule this interaction might be much more important due to direct connection of the two domains. This has implication for the quaternary structure of type VII collagen. The binding site does not overlap with potential collagen I binding areas. However, the mFNIII-9 mvWFA2 interaction might be influenced by binding of components of the extracellular matrix and thus the quaternary structure of type VII collagen can be affected. Due to the central role of Col7 in skin architecture disturbance of these structures either via mutations or autoantibodies leads to severe skin blistering diseases. These findings are the basis for further structural analysis and interaction studies with type I collagen and antibodies against mvWFA2. This will ultimately lead to a better understanding of pathogenesis of heritable and acquired skin blistering diseases. This work was supported by the cluster of excellence “inflammation at interfaces” of the DFG (EXC 306/1). The authors thank J. Brümmer for the BIAcore measurements and T. Peters and R. Ludwig for critical reading of the manuscript. The eNMR project (European FP7 e-Infrastructure grant, contract no. 213010, www.enmr.eu), supported by the national GRID Initiatives of Italy, Germany and the Dutch BiG Grid project (Netherlands Organization for Scientific Research), is acknowledged for the use of web portals, computing and storage facilities. Chemical shift data has been deposited in BioMagResBank under the accession number 17549 (http://www.bmrb.wisc.edu/). Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2011.04.071. Supplementary data. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.