A Predicted Atlas of Metal-Binding Sites Across the Protein Universe
Abstract
Metal-binding proteins (MBPs) play crucial roles in biological systems, including signal transduction, transcription, and catalysis. Recently, the AlphaFold Protein Structure Database (AFDB) has released approximately 214 million predicted protein structures, which provides a vast resource for the discovery of MBPs with novel structures and functions. However, systematic exploration of this massive dataset requires highly efficient computational approaches. Here, we present a machine learning model based on single-sequence input to predict metal-binding sites systematically across the whole AFDB protein universe. Using this approach, we established a comprehensive database termed “MetalDB” that contains approximately 38 million predicted MBPs with high confidence, and identified nearly 3 million sequence clusters that lack existing metal-binding annotations. Meta-analysis of MetalDB revealed a large number of metal-binding sites from previously unannotated protein families, structural domains as well as protein-protein interfaces, and uncovered metal-dependent functional proteins. Collectively, MetalDB provides a valuable resource for exploring MBPs across the whole protein universe and will significantly accelerate the discovery of proteins with novel functions in the post-genomic era.
References
1. Sendra, K. M. et al. An ancient metalloenzyme evolves through metal preference modulation. Nature Ecology & Evolution 7, 732–744 (2023).
2. Waldron, K. J., Rutherford, J. C., Ford, D. & Robinson, N. J. Metalloproteins and metal sensing. Nature 460, 823–830 (2009).
3. Holm, R. H., Kennepohl, P. & Solomon, E. I. Structural and Functional Aspects of Metal Sites in Biology. Chemical Reviews 96, 2239–2314 (1996).
4. Winkler, J. R. & Gray, H. B. Electron Flow through Metalloproteins. Chemical Reviews 114, 3369–3380 (2014).
5. Choi, T. S. & Tezcan, F. A. Overcoming universal restrictions on metal selectivity by protein design. Nature 603, 522–527 (2022).
6. Kim, D. et al. Computational design of metallohydrolases. Nature 649, 246–253 (2026).
7. Burger, N. et al. The human zinc-binding cysteine proteome. Cell 188, (2025).
8. Chen, L. et al. APEX2-based Proximity Labeling of Atox1 Identifies CRIP2 as a Nuclear Copper-binding Protein that Regulates Autophagy Activation. Angewandte Chemie International Edition 60, 25346–25355 (2021).
9. Bak, D. W. & Weerapana, E. Monitoring Fe–S cluster occupancy across the E. coli proteome using chemoproteomics. Nature Chemical Biology 19, 356–366 (2023).
10. Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Research 50, D439–D444 (2021).
11. Varadi, M. et al. AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences. Nucleic Acids Research 52, D368–D375 (2023).
12. Barrio-Hernandez, I. et al. Clustering predicted structures at the scale of the known protein universe. Nature 622, 637–645 (2023).
13. Lau, A. M. et al. Exploring structural diversity across the protein universe with The Encyclopedia of Domains. Science 386, eadq4946 (2024).
14. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
15. Hekkelman, M. L., de Vries, I., Joosten, R. P. & Perrakis, A. AlphaFill: enriching AlphaFold models with ligands and cofactors. Nature Methods 20, 205–213 (2023).
16. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 630, 493–500 (2024).
17. Krishna, R. et al. Generalized biomolecular modeling and design with RoseTTAFold All-Atom. Science 384, eadl2528 (2024).
18. Cheng, Y. et al. Co-evolution-based prediction of metal-binding sites in proteomes by machine learning. Nature Chemical Biology 19, 548–555 (2023).
19. Zhang, F. et al. MetalNet2: an enhanced server for predicting metal-binding sites in proteomes. National Science Review 11, nwae391 (2024).
20. Elnaggar, A. et al. Ankh: Optimized Protein Language Model Unlocks General-Purpose Modelling. arXiv preprint arXiv:2301.06568 (2023).
21. Team, B. A. A. et al. Protenix - Advancing Structure Prediction Through a Comprehensive AlphaFold3 Reproduction. bioRxiv https://doi.org/10.1101/2025.01.08.631967 (2025) doi:10.1101/2025.01.08.631967.
22. Mistry, J. et al. Pfam: The protein families database in 2021. Nucleic Acids Research 49, D412–D419 (2020).
23. Suzek, B. E. et al. UniRef clusters: a comprehensive and scalable alternative for improving sequence similarity searches. Bioinformatics 31, 926–932 (2014).
24. van Kempen, M. et al. Fast and accurate protein structure search with Foldseek. Nature Biotechnology 42, 243–246 (2024).
25. Youle, R. J. & Strasser, A. The BCL-2 protein family: opposing activities that mediate cell death. Nature Reviews Molecular Cell Biology 9, 47–59 (2008).
26. Chalkley, M. J., Mann, S. I. & DeGrado, W. F. De novo metalloprotein design. Nature Reviews Chemistry 6, 31–50 (2022).
27. Sillitoe, I. et al. CATH: increased structural coverage of functional space. Nucleic Acids Research 49, D266–D273 (2020).
28. Golan, G. et al. Structure of T4 Pyrimidine Dimer Glycosylase in a Reduced Imine Covalent Complex with Abasic Site-containing DNA. Journal of Molecular Biology 362, 241–258 (2006).
29. Bhopatkar, A. A., Uversky, V. N. & Rangachari, V. Chapter Nine - Disorder and cysteines in proteins: A design for orchestration of conformational see-saw and modulatory functions. in Dancing Protein Clouds: Intrinsically Disordered Proteins in Health and Disease, Part B (ed. Uversky, V. N.) vol. 174 331–373 (Academic Press, 2020).
30. Nugnes, M. V. et al. DisProt in 2026: enhancing intrinsically disordered proteins accessibility, deposition, and annotation. Nucleic Acids Research gkaf1175 (2025) doi:10.1093/nar/gkaf1175.
31. Levy, E. D. & Teichmann, S. A. Chapter Two - Structural, Evolutionary, and Assembly Principles of Protein Oligomerization. in Oligomerization in Health and Disease (eds Giraldo, J. & Ciruela, F.) vol. 117 25–51 (Academic Press, 2013).
32. Avraham, O., Tsaban, T., Ben-Aharon, Z., Tsaban, L. & Schueler-Furman, O. Protein language models can capture protein quaternary state. BMC Bioinformatics 24, 433 (2023).
33. Kshirsagar, M. et al. Rapid and accurate prediction of protein homo-oligomer symmetry using Seq2Symm. Nature Communications 16, 2017 (2025).
34. Ullanat, V., Jing, B., Sledzieski, S. & Berger, B. Learning the language of protein-protein interactions. Nature Communications 17, 1199 (2026).
35. Schweke, H. et al. An atlas of protein homo-oligomerization across domains of life. CELL 187, (2024).
36. Yamashita, E. et al. The host-binding domain of the P2 phage tail spike reveals a trimeric iron-binding structure. Acta crystallographica. Section F, Structural biology and crystallization communications 67, 837—841 (2011).
37. Zhang, R. et al. Structure of Escherichia coli Ribose-5-Phosphate Isomerase: A Ubiquitous Enzyme of the Pentose Phosphate Pathway and the Calvin Cycle. Structure 11, 31–42 (2003).
38. Tandogan, B. & Ulusu, N. N. Effects of cadmium and zinc ions on purified lamb kidney cortex glucose-6-phosphate dehydrogenase activity. Journal of Enzyme Inhibition and Medicinal Chemistry 21, 225–230 (2006).
39. Su, J. et al. SaProt: Protein Language Modeling with Structure-aware Vocabulary. in International Conference on Learning Representations (eds Kim, B. et al.) vol. 2024 6987–7009 (2024).
Metrics
DOI:
Submission ID:
Downloads
Posted
How to Cite
Download Citation
Declaration of Competing Interests
The authors declare no competing interests to disclose.
Copyright
The copyright holder for this preprint is the author/funder.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.