Large-scale crosslinking mass spectrometry (XL-MS) can generate low-resolution measurements of protein structure and protein-protein interaction (PPI) networks from complex samples [1]. For this, specialised protein crosslinkers, mass spectrometry (MS) approaches and/or database searching strategies are required. Here, we aimed to generate an ultra-deep XL-MS resource for human cells and investigate how these crosslinks can help inform and extend our understanding of the human proteome.
For two independent HEK293S cultures, we took a multi-pronged approach to diversify all aspects of crosslink generation and detection. Briefly, this included independent crosslinking of four subcellular fractions with three different crosslinker chemistries that react with lysine and/or acidic residues. Crosslinked peptides were enriched by offline size-exclusion chromatography and further fractionated by high pH reverse phase HPLC prior to LC-MSn. Spectra were acquired using hybrid-MS2-MS3 or high resolution MS/MS with EThcD or HCD fragmentation strategies. Crosslinked peptides were identified using XlinkX 2.3 or pLink2 at 1% FDR.
Our analysis generated the most comprehensive XL-MS dataset to date for any species, with 30551 unique residue-to-residue pairs (URPs) representing 4667 unique proteins and 3030 unique putative PPIs. This is more than double the number of URPs and crosslinked proteins compared to the most comprehensive human study to date. We show that these URPs confirm and rediscover existing experimental structures, capturing proteoforms and complexes within their approximate subcellular niches and range of conformations. Remarkably, our intra-molecular URPs also corroborate thousands of new structures predicted by next-generation modeller AlphaFold2, including those involving proteins (or regions of proteins) without existing resolution, and those lacking any structural precedent. Furthermore, our inter-protein crosslinks recapture the topology of well-described complexes and PPIs, whilst supporting or discovering poorly characterised PPIs. Critically, the inter-protein crosslinks also help localise PPI interfaces, and we use this information to assess quaternary protein structures modelled in AlphaFold-Multimer.