bioRxivBiophysicsPreprint
Full-Length Context Disrupts Folding of IgG-Binding Domains of Protein A
Kosar Rahimi, Albert Halbing, Minh Ngoc Nguyen, Mehmet ŞeniD, Richard C. WillsoniD, Gül H. ZerzeiD
Abstract
Multidomain proteins are often thought to fold as collections of independently stable domains, a modularity that underpins many assumptions in structural biology and design. Here, we challenge this view by examining the folding behavior of full-length Staphylococcal protein A (SpA), a 516-residue multidomain protein containing five immunoglobulin (Ig)-binding domains. Although each of the five Ig-binding domains of SpA folds stably in isolation (as it is already known experimentally and also confidently predicted by AI models), here, we show that the full-length construct and the individual Ig-binding domains in the full-length construct fail to adopt a stable three-dimensional structure in solution (despite being predicted to be folded by AI models). Instead, full-length SpA populates a compact yet predominantly disordered ensemble with residual secondary structure, where the folded state of each Ig-binding domain is thermodynamically unfavorable. These findings not only challenge long-held assumptions about the modular architecture and stability of SpA but also underscore the limitations of AI-based predictors when decoupled from the thermodynamic context. This work has implications for validating structure predictions, understanding multidomain architecture, and designing modular proteins for biotechnology and medicine.
- Citations
- 0
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Severe deviation in protein fold prediction by advanced AI: a case study
Jacinto López‐Sagaseta, Alejandro Urdiciain·Scientific ReportsPeer-reviewed·Feb 8, 2025
- AlphaFold3 and Intrinsically Disordered Proteins: Reliable Monomer Prediction, Unpredictable Multimer Performance
Tuan Minh Dao, Sebastien Ghent, … Taseef Rahman·Proteins Structure Function and BioinformaticsPeer-reviewed·Oct 12, 2026
- Structural analogue-based protein structure domain assembly assisted by deep learning
Chunxiang Peng, Xiaogen Zhou, … Guijun Zhang·BioinformaticsPeer-reviewed·Aug 13, 2022
- Benchmarking AI Protein Structure Predictors Reveals a Persistent Bias in Multi-State Proteins
Muhui Ye, Y Wang, … James C. Gumbart·bioRxiv·Jul 11, 2026
- Extending Conformational Ensemble Prediction to Multidomain Proteins and Protein Complex
Jun-Jie Zhu, Sören von Bülow, … Haifeng Chen·bioRxiv·Jan 15, 2026
- Accurate protein stability prediction for small domains using mega-scale experiments
Yehlin Cho, Kotaro Tsuboyama, … Gabriel J. Rocklin·bioRxiv·May 20, 2026