Peptide-centric analyses of human plasma enable increased resolution of biological insights into non-small cell lung cancer relative to protein-centric analysis

Abstract

Comprehensive assessment of the human proteome remains challenging due to multiple forms of a protein, or proteoforms, arising from alternative splicing, allelic variation, and protein modifications. As proteoforms can serve distinct functions and act as functional links between genotype and phenotype, proteoform-level knowledge is critical in understanding the molecular mechanisms underlying health and disease. However, identification of proteoforms requires unbiased protein coverage at amino acid resolution. Scalable, deep, and unbiased proteomics studies have been impractical due to cumbersome and lengthy workflows required for complex samples, like blood plasma. Here, we demonstrate the power of the Proteograph™ Product Suite in enabling unbiased, deep, and rapid proteomics at scale in a proof-of-concept proteoform analysis to dissect differences between protein isoforms in plasma samples from 80 healthy controls and 61 patients with early-stage non-small-cell lung cancer (NSCLC). Processing the 141 plasma samples with Proteograph yielded 22,993 peptides corresponding to 2,569 protein groups at a confidence of 1% false discovery rate. We extracted four proteins with peptides with significant abundance differences (p < 0.05; Benjamini-Hochberg corrected) in healthy control and cancer plasma samples. For one, the abundance variation can be explained by underlying annotated protein isoforms. For a second, we find evidence for differentially transcribed isoforms in the broader sequence data, but not in the known annotated protein isoforms. The others may be explained by novel isoforms or post-translational modifications. In addition, we sought to identify protein variants arising from allelic variation. To this end, we first performed whole exome sequencing on buffy coat samples from 29 individuals in the NSCLC study. Then, we created personalized mass spectrometry search databases for each individual subject from the exome sequences. From these libraries, we identified 422 protein variants, one of which has previously been shown to relate to lung cancer. In conclusion, our results demonstrate that Proteograph can generate unbiased and deep plasma proteome profiles that enable identification of proteoforms present in plasma at a scale sufficient to enable population-scale proteomic studies powered to reveal novel mechanistic and biomedical insights.