Proteome-Level Interplay between Folding and Aggregation Propensities of Proteins

Gian Gaetano Tartaglia, Michele VendruscoloJournal of Molecular Biology 2010 · DOI

TL;DR

This paper introduces novel computational methods (CamFold and Zyggregator) to predict protein folding and aggregation propensities directly from amino acid sequences. They reveal a fundamental interplay: regions that are crucial for protein folding often exhibit a high propensity for aggregation, suggesting a delicate balance that cellular machinery must manage.

What Problem Does This Paper Address?

Understanding how a protein folds into its functional 3D structure is a central challenge in molecular biology. Simultaneously, misfolding and aggregation of proteins are implicated in numerous diseases. While predicting protein structure from sequence has seen significant progress, predicting the kinetics of folding and the propensity for aggregation has remained more challenging. This paper aims to bridge this gap by developing sequence-based methods to predict both folding and aggregation propensities, and importantly, to investigate the relationship between these two processes at a proteome-wide level. The core question is: how do the inherent properties of a protein's amino acid sequence dictate its tendency to fold correctly versus misfold and aggregate?

Methods

The authors introduce two key computational methods:

  • CamFold: This method calculates a "folding propensity profile" for a protein sequence. It leverages physicochemical properties of amino acids (hydrophobicity, secondary-structure propensity, electrostatic charge) to assign a score to each residue, indicating its contribution to the overall folding process. The method incorporates local sequence context and is calibrated against experimental folding rates.
  • Zyggregator: This method, previously established by the authors, predicts the "aggregation propensity profile" of a protein sequence, also based on physicochemical properties.

Crucially, the paper distinguishes between "intrinsic aggregation propensity" (tendency to aggregate from the unfolded state) and "structure-corrected aggregation propensity" (tendency to aggregate from the folded state, where buried regions are protected). They also define "folding propensity profiles" in terms of "Zfold scores," which are normalized values derived from the CamFold method.

Datasets

  • Human and E. coli proteomes: Used for proteome-level analysis of folding and aggregation propensity profiles.
  • Database of 90 proteins with known folding rates: Used to train and validate the CamFold method's parameters.
  • Set of 250 E. coli proteins with experimentally determined GroEL/ES requirements: Used to investigate the relationship between folding/aggregation propensities and chaperone dependence.

Key Insights and Ideas

  • The Folding-Aggregation Trade-off: The most striking finding is the anticorrelation between folding propensity profiles and intrinsic aggregation propensity profiles. This suggests that regions critical for guiding protein folding often possess inherent characteristics that, if exposed, would promote aggregation.
  • Kinetic Control is Key: Proteins generally fold rather than aggregate under normal cellular conditions because the folding process is kinetically faster than aggregation. This allows aggregation-prone regions to become buried within the native structure before they can interact with other molecules.
  • Rate-Limiting Regions are Aggregation Hotspots: Regions identified as rate-limiting for folding (low Zfold scores) are often also identified as having high intrinsic aggregation propensity.
  • Secondary Structure Contributions: Beta-sheet regions tend to contribute most significantly to the Zfold scores, and negative Zfold contributions (indicating regions that slow down folding) are associated with high hydrophobicity and beta-sheet propensity, while also having high aggregation propensity.
  • Chaperone Dependence and Aggregation: Proteins with a higher dependence on molecular chaperones (like GroEL/ES) tend to exhibit a stronger anticorrelation between folding and aggregation propensities, suggesting that these proteins have regions that are particularly prone to aggregation and require assistance to fold correctly.

Key Results

  • Proteome-level analysis of human and E. coli showed a significant anticorrelation between folding and intrinsic aggregation propensity profiles.
  • The CamFold method achieved a correlation coefficient of 0.87 with experimentally measured folding rates for a diverse set of proteins, demonstrating its predictive power.
  • Regions with low folding propensity (rate-limiting for folding) were found to have low disorder propensity and high aggregation propensity.
  • Proteins classified as highly dependent on chaperones (classes II and III) showed a more pronounced anticorrelation between folding and aggregation propensity profiles compared to those less dependent (class I).

Limitations and Future Work

  • The authors note that the current implementation of CamFold is not biased towards specific protein classes extensively studied in literature but may require modifications to predict the effects of amino acid mutations.
  • They suggest that the CamFold method could be used to predict folding rates at different pH values, as the underlying scales are pH-dependent.
  • The study highlights the importance of understanding the interplay between folding and aggregation for a comprehensive view of protein behavior, with implications for both fundamental biology and disease.

Summary updated

BibTeX
@article{tartaglia2010proteome,
  author = {Gian Gaetano Tartaglia and Michele Vendruscolo},
  title = {{Proteome-Level Interplay between Folding and Aggregation Propensities of Proteins}},
  journal = {Journal of Molecular Biology},
  year = {2010},
  doi = {10.1016/j.jmb.2010.08.013},
}
Figure