This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.
Application Note pubs.acs.org/jcim
MIB: Metal Ion-Binding Site Prediction and Docking Server Yu-Feng Lin,† Chih-Wen Cheng,† Chung-Shiuan Shih,† Jenn-Kang Hwang,† Chin-Sheng Yu,‡,§ and Chih-Hao Lu*,⊥ †
Institute of Bioinformatics and Systems Biology, National Chiao Tung University, Hsinchu, 30050 Taiwan Department of Information Engineering and Computer Science and §Master’s Program in Biomedical Informatics and Biomedical Engineering, Feng Chia University, Taichung 40724, Taiwan ⊥ Graduate Institute of Basic Medical Science, China Medical University, Taichung 40402, Taiwan ‡
ABSTRACT: The structure of a protein determines its biological function(s) and its interactions with other factors; the binding regions tend to be conserved in sequence and structure, and the interacting residues involved are usually in close 3D space. The Protein Data Bank currently contains more than 110 000 protein structures, approximately onethird of which contain metal ions. Identifying and characterizing metal ion-binding sites is thus essential for investigating a protein’s function(s) and interactions. However, experimental approaches are time-consuming and costly. The web server reported here was built to predict metal ion-binding residues and to generate the predicted metal ion-bound 3D structure. Binding templates have been constructed for regions that bind 12 types of metal ion-binding residues have been used to construct binding templates. The templates include residues within 3.5 Å of the metal ion, and the fragment transformation method was used for structural comparison between query proteins and templates without any data training. Through the adjustment of scoring functions, which are based on the similarity of structure and binding residues. Twelve kinds of metal ions (Ca2+, Cu2+, Fe3+, Mg2+, Mn2+, Zn2+, Cd2+, Fe2+, Ni2+, Hg2+, Co2+, and Cu+) binding residues prediction are supported. MIB also provides the metal ions docking after prediction. The MIB server is available at http://bioinfo.cmu.edu.tw/MIB/.
■
INTRODUCTION There are many different types of metal ion-binding proteins, and those that bind the most common metal ions, such as iron, usually regulate essential functions in physiological processes.1−3 However, the functions of many metal ion-binding proteins remain unclear. Therefore, identification of the positioning of metal-binding residues in 3D space is important for clarifying the ion specificity of a protein and will provide general insight into the possible roles of metal ions in protein function.
Protein structure determines biophysical functions and interactions with other components, including metal ions, small ligands, and other proteins.4,5 Metal ions can stabilize protein structures and participate in catalysis; therefore identifying metal ion-binding sites is key to understanding the biological relevance of metal ion-binding proteins. Identifying metal-binding sites experimentally can be difficult and tedious because the process requires complicated steps or specialized techniques, such as Table 2. Predictive Performance of MIB at a Greater than 95% Specificity Threshold
Table 1. Types and Numbers of Metal Ion-Binding Polypeptide Chains Examined and Metal Ion-Binding Residue Templates Established metal ion
number of polypeptide chains
number of templates
Ca2+ Cu2+ Fe3+ Mg2+ Mn2+ Zn2+ Cd2+ Fe2+ Ni2+ Hg2+ Co2+ Cu+ total
273 47 51 256 110 372 45 82 31 32 151 46 1496
407 74 77 209 144 499 110 101 39 50 209 78 1997
© 2016 American Chemical Society
metal ion
accuracy (%)
sensitivity (%)
Ca2+ Cu2+ Fe3+ Mg2+ Mn2+ Zn2+ Cd2+ Fe2+ Ni2+ Hg2+ Co2+ Cu+
94.1 94.9 94.9 94.6 95.0 94.8 92.9 95.1 94.7 94.1 94.7 95.3
48.9 85.6 85.4 37.0 61.4 71.1 41.2 92.9 70.7 40.9 64.5 76.6
Received: July 14, 2016 Published: November 29, 2016 2287
DOI: 10.1021/acs.jcim.6b00407 J. Chem. Inf. Model. 2016, 56, 2287−2291
Application Note
Journal of Chemical Information and Modeling nuclear magnetic resonance spectroscopy,6 gel electrophoresis,7 metal-affinity column chromatography and electrophoretic mobility shift assays,8 and absorption spectroscopy.9 In contrast, development of computational methods may allow relatively easy and fast analysis of metal-binding sites. This approach can take advantage of the large and accessible database of protein
structures in the Protein Data Bank (PDB),10 which currently contains over 110 000 protein structures and is continually growing, to develop prediction tools. Multiple diverse tools have been developed to examine different aspects of protein interactions, such as QUARK,11 which predicts protein structures, and GRID,12 COACH,13 Bspred,14 CHED,15 SeqCHED,16 and Metaldetector,17 which predict ligand-binding sites. However, few tools have been developed for the prediction of diverse metal-binding sites and the docking of metal ion. We previously used the fragment transformation method18 to predict binding sites for six metal ions, namely Ca2+, Cu2+, Fe3+, Mg2+, Mn2+, and Zn2+. Here, we also analyzed additional six metal ion-binding sites which were Cd2+, Fe2+, Ni2+, Hg2+, Co2+, and Cu+. MIB (metal ion binding site prediction, http://bioinfo.cmu.edu.tw/MIB/) was established for not only 12 metal-binding sites prediction but also metal ion docking. Our method was simple without any data training for prediction and without complicated force field calculation for docking. MIB was a comprehensive and userfriendly web tool for reliable prediction of metal ion-binding sites. By assisting in the elucidation of metal-binding sites in proteins, MIB could potentially enrich the output of proteomics studies.
■
METHODS AND IMPLEMENTATION Overview. MIB is a binding site prediction and docking server for metal ions, and this server provides an accurate, integrated approach to search the residues in metal ion-binding sites using the fragment transformation method.18 Predictions of residues that bind 12 types of metal ions are supported. The query protein structure is compared with each metal-binding template in the database to locate the metal-binding residues. Each residue of the query protein is assigned a binding score,
Figure 1. Example of the MIB input page http://bioinfo.cmu.edu.tw/ MIB/ (accessed Nov 18, 2016). Twelve types of metal ion can be chosen for binding site prediction.
Figure 2. Prediction window of the MIB output page http://bioinfo.cmu.edu.tw/MIB/structure/f82eaf7925500b61991848fbc7f4aec7_ (accessed Nov 18, 2016). The predicted metal ion-binding residues are indicated by a blue label and shown as sticks in the in the 3D representation. 2288
DOI: 10.1021/acs.jcim.6b00407 J. Chem. Inf. Model. 2016, 56, 2287−2291
Application Note
Journal of Chemical Information and Modeling
Figure 3. Docking window of the MIB output page http://bioinfo.cmu.edu.tw/MIB/structure/f82eaf7925500b61991848fbc7f4aec7_ (accessed Nov 18, 2016). The predicted docking position of the metal ion is shown in the 3D representation, and the curve of predicted scores for the residues of whole query protein is shown at the page bottom.
which is composed of sequence and structure conservation measures. When the binding score of a residue is higher than a specified threshold, this residue is predicted to be a metalbinding residue. Based on the local 3D structure alignment between query protein and metal ion-binding template, the metal ion in the metal-binding template can be transformed into the query protein structure. MIB also provides metal iondocking visualization functions after prediction. Metal Ion-Binding Residue Templates Database. The structures of protein complexes containing at least one Ca2+, Cu2+, Fe3+, Mg2+, Mn2+, Zn2+, Cd2+, Fe2+, Ni2+, Hg2+, Co2+, or Cu+ ion were collected from the PDB. Homologous proteins were filtering out, and the remaining proteins had pairwise sequence identity 95%, our method yielded an accuracy from 92.9 to 95.1%, and for 8 of the 12 ions the sensitivity was >60%. For binding sites of Cu2+, Fe3+, and Fe2+, prediction sensitivity >85% was achieved. Application. A 3D protein structure is necessary as the target input for MIB prediction, and it can be uploaded as a protein structure in PDB file format or downloaded from the PDB by entering the PDB ID. The specific chain of interest in the target protein structure and the metal ion type both need to be selected when submitting the prediction request. Twelve types of metal ion can be chosen for binding site prediction (Figure 1). The results of an inquiry are shown in two windows, “Prediction” (Figure 2) and “Docking” (Figure 3). In the prediction window, the binding scores of all residues are listed in the order that the residues appear in the sequence, with the predicted metal ion-binding residues indicated by blue label. A manipulatable stick diagram of the predicted binding residues can be shown in the prediction window or be hidden. In addition to predicting binding residues and estimating the binding scores, MIB also predicts the docked position of the metal ion in the protein structure. In the docking window, the aligned templates are listed in a table. By clicking the visualization icon of a specific template, this template structure will be shown above the table, and the corresponding aligned residues and predicted metal ion will also be shown in the query protein structure. Further, 2290
DOI: 10.1021/acs.jcim.6b00407 J. Chem. Inf. Model. 2016, 56, 2287−2291
Application Note
Journal of Chemical Information and Modeling (12) Wright, B.; Watson, K. A.; McGuffin, L. J.; Lovegrove, J. A.; Gibbins, J. M. GRID and docking analyses reveal a molecular basis for flavonoid inhibition of Src family kinase activity. J. Nutr. Biochem. 2015, 26, 1156. (13) Yang, J.; Roy, A.; Zhang, Y. Protein-ligand binding site recognition using complementary binding-specific substructure comparison and sequence profile alignment. Bioinformatics 2013, 29, 2588−95. (14) Mukherjee, S.; Zhang, Y. Protein-protein complex structure predictions by multimeric threading and template recombination. Structure 2011, 19, 955−66. (15) Babor, M.; Gerzon, S.; Raveh, B.; Sobolev, V.; Edelman, M. Prediction of transition metal-binding sites from apo protein structures. Proteins: Struct., Funct., Genet. 2008, 70, 208−17. (16) Levy, R.; Edelman, M.; Sobolev, V. Prediction of 3D metal binding sites from translated gene sequences based on remotehomology templates. Proteins: Struct., Funct., Genet. 2009, 76, 365−74. (17) Passerini, A.; Lippi, M.; Frasconi, P. MetalDetector v2.0: predicting the geometry of metal binding sites from protein sequence. Nucleic Acids Res. 2011, 39, W288−92. (18) Lu, C. H.; Lin, Y. S.; Chen, Y. C.; Yu, C. S.; Chang, S. Y.; Hwang, J. K. The fragment transformation method to detect the protein structural motifs. Proteins: Struct., Funct., Genet. 2006, 63, 636− 43. (19) Henikoff, S.; Henikoff, J. G. Amino acid substitution matrices from protein blocks. Proc. Natl. Acad. Sci. U. S. A. 1992, 89, 10915−9.
2291
DOI: 10.1021/acs.jcim.6b00407 J. Chem. Inf. Model. 2016, 56, 2287−2291
