Improved Interpretation of Protein Conformational Differences and Ligand Occupancy in Large-Scale Cross-Link Data

Chemical
cross-linking of proteins in complex samples, cells, or
even tissues is emerging to provide unique structural information
on proteins and complexes that exist within native or nativelike environments.
The public database XLinkDB automatically maps cross-links to available
structures based on sequence homology. Structures most likely to reflect
protein conformations in the cross-linked sample are routinely identified
by having cross-linked residues separated by Euclidean distances within
the maximum span of the applied cross-linker. Solvent accessible surface
distance (SASD), which considers the accessibility of the cross-linked
residues and the path connecting them, is a better predictor of consistency
than the Euclidean distance. However, SASDs of structures are not
publicly available, and their calculation is computationally intensive.
Here, we describe in XLinkDB version 4.0 the automatic calculation
of SASDs using Jwalk for all cross-links mapped to structures, both
with and without regard to ligands, and derive empirical maximum SASD
spans for BDP-NHP and DSSO cross-linkers of 51 and 43 Å, respectively.
We document ligands proximal to cross-links in structures and demonstrate
how SASDs can be used to help infer sample protein conformations and
ligand occupancy, highlighting cross-links sensitive to ADP binding
in mitochondria isolated from HEK293 cells.