TL;DR
This paper introduces "BDamage," a novel atomic metric to quantify site-specific radiation damage in protein crystal structures. By analyzing a large dataset of PDB structures, the authors demonstrate that BDamage can identify regions of a protein susceptible to damage and reveal correlations between damage susceptibility and physicochemical properties like solvent accessibility and disulfide bond conformation, offering a new tool for understanding and mitigating radiation damage in structural biology.
What Problem Does This Paper Address?
X-ray crystallography is a powerful technique for determining the 3D structure of proteins, but it's inherently damaging to the samples. While global radiation damage (affecting the entire diffraction pattern) is well-known, the paper focuses on site-specific radiation damage. This type of damage affects particular parts of a protein in a predictable, repeatable way, and can lead to incorrect interpretations of protein function and mechanism. For example, disulfide bonds are known to break early, and certain amino acids like aspartic and glutamic acids can decarboxylate. However, this damage doesn't occur uniformly even within the same type of structural motif; some regions are more susceptible than others, a phenomenon termed "preferential specific damage." Existing methods for assessing this damage are often indirect or require complex analysis of diffraction data. The authors aim to provide a simple, atomic-level metric that can be calculated directly from a refined PDB structure to identify and quantify these damage hotspots.
Methods
The core of the paper is the introduction of a new metric called BDamage. This metric is designed to quantify site-specific radiation damage at the atomic level. Here's how it works:
- Atomic B-factors: The metric starts with atomic B-factors (also known as atomic displacement parameters) from a PDB file. These values represent the uncertainty in an atom's position, influenced by thermal motion and crystalline disorder. While high B-factors can indicate disorder, they don't directly tell us about radiation damage.
- Packing Density: To account for the fact that B-factors are influenced by how tightly an atom is packed, the authors calculate a "packing density" for each atom. This is done using atomic contact numbers (ACNs), considering neighboring atoms within a defined radius (14 Å in their primary analysis), and importantly, including symmetry-related copies of the protein in the crystal.
- BDamage Calculation: For each atom, its B-factor is normalized by the average B-factor of all other non-hydrogen atoms within the same packing density environment. This normalization is key: it aims to isolate the damage-induced disorder from the inherent disorder due to packing. A BDamage value greater than 1 suggests that an atom is more disordered than expected for its packing environment, potentially indicating radiation damage.
The authors then validate this metric using two main datasets:
- A reference dataset of six protein crystal pairs (Nanao et al., 2005), where one dataset was collected before an X-ray "burn" (low dose) and the other after (high dose). This allows for direct comparison of damage levels.
- A large PDB survey of over 2700 protein structures to investigate broader correlations between BDamage and various physicochemical properties.
Datasets
- Nanao et al. (2005) dataset: This dataset consists of six pairs of protein crystal structures. For each protein, two diffraction datasets were collected from the same crystal: one before an X-ray "burn" (low dose) and one after a significant X-ray exposure (high dose). This dataset was crucial for validating the BDamage metric by directly comparing low-dose and high-dose states.
- Large PDB subset: A dataset of 2704 non-redundant protein structures was compiled from the Protein Data Bank (PDB). These structures met specific criteria for resolution (1.5-1.8 Å) and R-factor (≤ 20%). This large dataset was used to perform statistical analyses and identify general trends and correlations related to BDamage.
Key Insights and Ideas
- BDamage as a direct damage indicator: The most significant insight is that BDamage provides a more sensitive and direct measure of site-specific radiation damage than raw atomic B-factors. It effectively highlights regions that are disproportionately disordered due to radiation, even when overall B-factors don't show a clear trend.
- Preferential damage is real and quantifiable: The paper confirms that radiation damage is not uniform. BDamage reveals that certain residues and even specific atoms within those residues are significantly more susceptible to damage than others within the same protein.
- Solvent accessibility is a key predictor: A strong, consistent positive correlation was found between BDamage and solvent accessibility. This suggests that more exposed regions of a protein are more vulnerable to radiation damage.
- Disulfide bond conformation matters: Different types of disulfide bonds (spiral, hook, staple) exhibit distinct BDamage distributions, indicating varying susceptibilities. This is particularly interesting as staple-group disulfide bonds are often associated with protein function regulation.
- Secondary structure has minimal impact: Contrary to what one might expect, the study found that BDamage is largely independent of secondary protein structure (e.g., alpha-helices, beta-sheets), suggesting that damage is more influenced by local chemical environment and accessibility.
Key Results
- The BDamage metric successfully distinguished between low-dose and high-dose datasets in the Nanao et al. reference set, showing a marked increase in BDamage for susceptible atoms (like cysteine sulfurs and carboxylate termini) in the high-dose samples.
- Statistical analysis of the large PDB subset revealed a consistent positive correlation between BDamage and solvent accessibility for various susceptible amino acid residues (ASP, CYS, GLU, MET, TYR).
- BDamage distributions differed significantly between different disulfide bond groups (spiral, hook, staple), with staple-group disulfide bonds showing a tendency towards higher BDamage values.
- The study found that BDamage is largely independent of secondary protein structure, with the exception of π-helix regions, which may be due to sampling effects.
Limitations and Future Work
- PDB quality dependence: The BDamage metric relies on high-quality PDB structure models with well-refined atomic B-factors. It cannot currently handle multiple models or partial occupancies, which can be common in PDB entries.
- Unmodeled regions: Regions of the protein that are not well-modeled in the PDB file can affect both the average B-factor and packing density calculations, potentially impacting BDamage accuracy.
- Anisotropic B-factors: The current BDamage metric does not utilize anisotropic atomic B-factors, which provide more detailed information about atomic displacement.
- Future Work: The authors suggest that a protein-wide variant of BDamage could serve as an overall quality indicator for structural damage. They also propose that BDamage could be used during structure refinement and deposition to identify candidate locations for radiation-induced structural changes and to ensure the accuracy of published models. Further investigation into the exact level of radiation dose at which BDamage becomes a reliable indicator is also suggested.
Summary updated