TL;DR
This study investigates why stabilizing mutations can paradoxically decrease protein solubility and increase aggregation. By combining experimental and computational methods, the researchers found that for one enzyme (LinB116), stabilization exposed "cryptic aggregation-prone regions" that were previously buried, leading to aggregation during unfolding. This highlights the importance of considering unfolding pathways and hidden vulnerabilities when engineering proteins.
What Problem Does This Paper Address?
Proteins are crucial for biotechnology, but natural proteins often lack the stability or solubility needed for industrial applications. While protein engineering can enhance stability, it sometimes comes at the cost of reduced solubility and increased aggregation. This phenomenon, where stabilizing mutations lead to aggregation, is counterintuitive and poorly understood. The paper specifically addresses this by investigating two highly homologous haloalkane dehalogenases (HLDs): DhaA and LinB. Despite their structural similarity, LinB is known for its poor solubility and aggregation issues, which are exacerbated upon stabilization. The central question is: what are the molecular mechanisms behind this trade-off between stability and solubility, particularly in the context of protein unfolding?
Methods
The researchers employed a multi-pronged approach:
- Experimental Techniques:
- Differential Scanning Fluorimetry (DSF) and Differential Scanning Calorimetry (DSC): Used to measure protein melting temperatures (stability) and assess unfolding reversibility.
- Static Light Scattering (SLS): Employed to detect and quantify protein aggregation.
- Circular Dichroism (CD) Spectroscopy: Used to monitor changes in protein secondary structure during unfolding.
- Computational Techniques:
- Molecular Dynamics (MD) Simulations: Used to simulate the unfolding process of the proteins at high temperatures (98°C) to accelerate sampling.
- Markov State Models (MSMs): Constructed from MD trajectories to identify distinct conformational states (folded, intermediate, unfolded) and their transitions.
- Adaptive Sampling: A technique to efficiently explore the conformational space of the proteins during unfolding simulations.
- Aggregation Propensity Prediction Tools (AggreProt, Aggrescan3D 2.0): Used to identify regions within the protein sequence that are prone to aggregation.
- Homology Modeling (SWISS-MODEL): Used to generate the structure of the stabilized LinB116 mutant.
The integration of these methods allowed for a comprehensive analysis of unfolding pathways, conformational changes, and the identification of aggregation-prone regions, both in the native state and during unfolding.
Datasets
The study utilized experimental data from four protein variants: DhaAwt, DhaA115, LinBwt, and LinB116. The structures of DhaAwt, DhaA115, and LinBwt were obtained from the RCSB Protein Data Bank. The structure for LinB116 was generated via homology modeling. The computational simulations generated extensive trajectory data, which was then used to construct Markov State Models.
Key Insights and Ideas
- Unfolding Pathways Dictate Aggregation: The study revealed that the way a protein unfolds is critical for its aggregation behavior. LinB116, unlike DhaA115, aggregates more readily because its unfolding pathway leads to the exposure of "cryptic aggregation-prone regions" (APRs) that are normally buried in the folded state.
- "Cryptic APRs" are the Culprits: The key insight is the identification of these hidden APRs. These are regions that are not prone to aggregation in the native state but become exposed during unfolding, particularly in the intermediate states. For LinB116, stabilization mutations seem to have inadvertently "unlocked" these cryptic APRs.
- Hydrophobicity Trade-off: Stabilizing mutations often introduce hydrophobic residues to improve internal packing and stability. However, for LinB116, this increased surface hydrophobicity contributed to its poor solubility and aggregation propensity.
- DhaA vs. LinB: Different Strategies: DhaA variants, even when stabilized, tend to expose residues more broadly and less specifically related to APRs. LinB variants, on the other hand, show a more localized exposure of residues within specific APRs during unfolding, especially LinB116.
- Flexibility vs. Rigidity in Stabilization: The study suggests that DhaA115 stabilization might involve increased entropy (flexibility), while LinB116 stabilization might involve rigidification. This highlights that different mechanisms can lead to enhanced stability, with varying consequences for other properties.
Key Results
- LinB116 Aggregation is Unfolding-Driven: Experimental data showed that LinB116 aggregation onset often preceded or coincided with unfolding, unlike LinBwt where it followed unfolding.
- MD Simulations Identified Key Unfolding Regions: Molecular dynamics simulations identified specific regions (cap domain, loops L5, L9, L16, L18, and helix α9) that were highly flexible and underwent significant conformational changes during unfolding for all variants.
- Cryptic APRs in LinB: Analysis of exposed residues during unfolding, combined with aggregation propensity predictions, revealed that LinB variants, particularly LinB116, expose residues within predicted APRs more significantly than DhaA variants. Specifically, 60% of exposed residues in LinB116 during unfolding were located within previously identified APRs.
- Increased Surface Hydrophobicity in LinB116: Stabilizing mutations in LinB116 led to a more pronounced increase in surface hydrophobicity (4% and 21% depending on the scale) compared to DhaA115 (0.1% and 4%).
- Engineering Cryptic APRs Shows Promise: Attempts to engineer these cryptic APRs in LinB116, by introducing hydrophilic residues, showed some promise in reducing aggregation, although it also led to destabilization.
Limitations and Future Work
- MD Simulation Temperature: The MD simulations were conducted at 98°C, which is above the melting temperature. While the authors state this doesn't alter the unfolding pathway, it's a high-temperature condition that might not perfectly reflect physiological unfolding.
- Computational Demands: The authors acknowledge that MD sampling of unfolded and partially unfolded conformations is computationally demanding and a bottleneck for this approach.
- Complexity of Aggregation: The study suggests that LinB aggregation might involve at least two independent pathways, with the cryptic APR pathway being a significant contributor to irreversibility.
- Future Directions: The authors propose that their methodology for identifying cryptic APRs can be applied to other proteins. They also highlight the need for more efficient and low-resource methods for generating conformational ensembles. Advancements in machine learning for protein dynamics are seen as key to making this approach more practical for routine predictions. They also suggest further engineering of these cryptic regions in LinB to improve solubility, while carefully balancing stability.
Summary updated