Structure of a Functional Amyloid Protein Subunit Computed Using Sequence Variation

Pengfei Tian, Wouter Boomsma, Yong Wang, Daniel E. Otzen, Mogens H. Jensen, Kresten Lindorff‐LarsenJournal of the American Chemical Society 2014 · DOI

TL;DR

Researchers have computationally predicted the 3D structure of CsgA, a key protein subunit of functional amyloid fibers called curli, by analyzing evolutionary sequence variations. This novel approach, combining coevolutionary analysis with molecular simulations, reveals that CsgA forms a β-helical structure, offering new insights into the assembly and function of these important bacterial structures.

What Problem Does This Paper Address?

Functional amyloids, like bacterial curli fibers, play crucial roles in biological processes, contrasting with their disease-associated counterparts. However, their insoluble and non-crystalline nature makes traditional structural determination methods (like X-ray crystallography or solution NMR) challenging. This paper tackles the problem of determining the atomic-resolution structure of CsgA, the primary subunit of curli fibers, which are essential for bacterial adhesion and invasion. Understanding CsgA's structure is key to comprehending curli assembly and function, and potentially to developing strategies to modulate their activity.

Methods

The core of the methodology lies in leveraging coevolutionary analysis of amino acid sequences. The authors constructed a multiple sequence alignment (MSA) of CsgA homologues and used two distinct methods, EVCOUPLING and PSICOV, to infer residue-residue contacts. These predicted contacts, representing amino acids that are spatially close in the folded protein, were then used as structural restraints in molecular simulations. A hybrid pseudoenergy function was employed, combining the coevolutionary contact potential with an efficient all-atom implicit solvent force field (ProFASi). Monte Carlo simulations were initiated from fully extended structures to explore conformational space and identify low-energy states.

Datasets

The primary dataset used was a multiple sequence alignment (MSA) of 390 CsgA homologues, obtained from the HHblits2 and uniprot20 databases. No experimentally solved structures were available for these homologues, necessitating the de novo structure prediction approach. The authors also mention using experimental data (solid-state NMR, electron microscopy, X-ray diffraction) for validation, and they tested their coevolutionary prediction method on a protein with a known β-helical structure to demonstrate its accuracy.

Key Insights and Ideas

  • Evolutionary covariation as a structural probe: The paper demonstrates that analyzing patterns of amino acid changes across related protein sequences can reveal crucial information about their 3D structure, even for proteins that are difficult to crystallize.
  • CsgA adopts a β-helical structure: Contrary to expectations, the simulations consistently predict a stable β-helical fold for CsgA, a structure known for its stability due to inter-strand hydrogen bonds and a hydrophobic core.
  • Left-handed vs. Right-handed β-helices: Interestingly, the simulations produced both left-handed and right-handed β-helical structures with similar energies, suggesting that the evolutionary pressure might allow for either orientation, or that the fibril environment dictates the final handedness.
  • Repeat sequences map to β-helix turns: The previously identified repeat sequences within CsgA align with the turns of the β-helical structure, providing a compelling link between sequence organization and predicted fold.
  • Functional amyloids are evolutionarily selected for structure: The fact that CsgA has evolved to form a stable amyloid structure implies that this fold is beneficial and has been maintained by natural selection, highlighting the "designed" nature of functional amyloids.

Key Results

  • The coevolutionary analysis revealed a striking pattern of contacts parallel to the diagonal in the sequence alignment, strongly suggesting an internally homologous and potentially helical structure.
  • Molecular simulations using these contacts and the ProFASi force field consistently converged to β-helical structures, both left-handed and right-handed.
  • The predicted β-helical structures were validated against experimental data:
  • X-ray fiber diffraction: Atomic spacings of ~4.7 and ~9 Å matched the predicted inter-strand and inter-sheet distances.
  • Electron microscopy: The predicted fibril width (~30-31 Å) agreed with experimental observations.
  • Solid-state NMR (ssNMR): Predicted inter-residue distances for Val, Leu, and Phe were in good agreement with experimental measurements (~6-8 Å). Calculated backbone chemical shifts also showed good agreement with experimental ssNMR data.

Limitations and Future Work

  • Ambiguity in handedness: The simulations produced both left- and right-handed β-helices with similar energies, and the available experimental data (NMR chemical shifts, inter-residue distances) could not definitively distinguish between them. The authors suggest that one or both might represent the structural unit, or that the fibril environment dictates the final handedness.
  • Monomer vs. Fibril: The current study focuses on the monomeric structure. While the predicted monomer structure is validated, the interactions between CsgA monomers to form the complete fibril remain difficult to study with the current approach, as no strong sequence signals for inter-monomer contacts were found.
  • Future work: The authors suggest further experiments to measure inter-residue distances for residues like Ile and Gln to better probe the conformation. They also envision that with growing sequence databases and improved methods, the approach could be extended to study the organization of entire fibrils.

Summary updated

Abstract

Functional amyloid fibers, called curli, play a critical role in adhesion and invasion of many bacteria. Unlike pathological amyloids, curli structures are formed by polypeptide sequences whose amyloid structure has been selected for during evolution. This important distinction provides us with an opportunity to obtain structural insights from an unexpected source: the covariation of amino acids in sequences of different curli proteins. We used recently developed methods to extract amino acid contacts from a multiple sequence alignment of homologues of the curli subunit protein, CsgA. Together with an efficient force field, these contacts allow us to determine structural models of CsgA. We find that CsgA forms a β-helical structure, where each turn corresponds to previously identified repeat sequences in CsgA. The proposed structure is validated by previously measured solid-state NMR, electron microscopy, and X-ray diffraction data and agrees with an earlier proposed model derived by complementary means.
BibTeX
@article{tian2014structure,
  author = {Pengfei Tian and Wouter Boomsma and Yong Wang and Daniel E. Otzen and Mogens H. Jensen and Kresten Lindorff‐Larsen},
  title = {{Structure of a Functional Amyloid Protein Subunit Computed Using Sequence Variation}},
  journal = {Journal of the American Chemical Society},
  year = {2014},
  doi = {10.1021/ja5093634},
}
Figure