Development
of new reagents for protein cross-linking is constantly
ongoing. The chemical formulas for the linker adducts formed by these
reagents are usually deduced from expert knowledge and then validated
by mass spectrometry. Clearly, it would be more rigorous to infer
the chemical compositions of the adducts directly from the data without
any prior assumptions on their chemistries. Unfortunately, the analysis
tools that are currently available to detect chemical modifications
on linear peptides are not applicable to the case of two cross-linked
peptides. Here, we show that an adaptation of the open search strategy
that works on linear peptides can be used to characterize cross-link
modifications in pairs of peptides. We benchmark our approach by correctly
inferring the linker masses of two well-known reagents, DSS and formaldehyde,
to accuracies of a few parts per million. We then investigate the
cross-linking chemistries of two poorly characterized reagents: EMCS
and glutaraldehyde. In the case of EMCS, we find that the expected
cross-linking chemistry is accompanied by a competing chemistry that
targets other amino acid types. In the case of glutaraldehyde, we
find that the chemical formula of the dominant linker is C
5
H
4
, which indicates a ringed aromatic structure. These
results demonstrate how, with very little effort, our approach can
yield nontrivial insights to better characterize new cross-linkers.