Cross-platform proteomics signatures of extreme old age

Abstract

In previous work, we used a SomaLogic platform targeting approximately 5000 proteins to generate a serum protein signature of centenarians that we validated in independent studies that used the same technology. We set here to validate and possibly expand the results by profiling the serum proteome of a subset of individuals included in the original study using liquid chromatography tandem mass spectrometry (LC-MS/MS). Following pre-processing, the LC-MS/MS data provided quantification of 398 proteins, with only 266 proteins shared by both platforms. At 1% FDR statistical significance threshold, the analysis of LC-MS/MS data detected 44 proteins associated with extreme old age, including 23 of the original analysis. To identify proteins for which associations between expression and extreme-old age were conserved across platforms, we performed inter-study conservation testing of the 266 proteins quantified by both platforms using a method that accounts for the correlation between the results. From these tests, a total of 80 proteins reached 5% FDR statistical significance, and 26 of these proteins had concordant pattern of gene expression in whole blood generated in an independent set. This signature of 80 proteins points to blood coagulation, IGF signaling, extracellular matrix (ECM) organization, and complement cascade as important pathways whose protein level changes provide evidence for age-related adjustments that distinguish centenarians from younger individuals. The comparison with blood transcriptomics also highlights a possible role for neutrophil degranulation in aging.

Keywords: Extreme longevity; Mass spectrometry; Serum proteomics; SomaLogic.

Fig. 1

Inter-study protein signatures comparing centenarians to offspring and unrelated controls. A UpSet plot of protein signatures detected in the labelled LC-MS/MS dataset with 50 subjects (Mass Spec), the SomaScan subset with the same 50 subjects (SomaScan Subset), and the SomaScan data with 224 participants (Somascan) at a 1% FDR significance threshold. Numbers in parentheses with a “*” denote the proteins that were included in both datasets but detected only in one analysis. Note that five proteins were identified with “Conserved” signal that did not reach statistical significance in any individual study. B Fisher’s exact test results of the intersection between the labelled LC-MS/MS and SomaScan signatures. The diagonal represents the number of significant proteins from each analysis that are annotated by the SomaScan assay. C Comparison of log2 fold changes protein expression differences of controls and offspring to centenarians from LC-MS/MS and SomaScan studies across their 266 shared proteins. D Log2 fold changes protein expression differences inter-study conserved proteins, including each LC-MS/MS result and SomaScan aptamer. All protein signature results shown were identified using an FDR cutoff of 0.01. The full set of analysis results are reported in Supplementary Table S2

Fig. 2

Protein interaction and functional analysis results. These results reflect analyses of 96 protein signature of proteins with conserved (FDR values < 0.05) and labelled LC-MS/MS-only discovered proteins (FDR < 0.01). A Overrepresentation-based enrichment analysis comparing Gene Ontology Functional Terms and Reactome pathways to three protein lists: All 96 proteins (All), 44 proteins with higher expression in offspring and controls (Up in Off./Cont.) and 52 proteins with higher expression in centenarians (up in centenarians). Additional information for these results, including term source, p-values, set sizes, and set members are reported in Supplementary Table S3. B STRING database annotated protein physical and functional interactions. The full list of interaction pairs is reported in Supplementary Table S4

Fig. 3

Summary of literature and Reactome database confirmed protein physical interactions. Additional information, including descriptions of interaction and sources can be found in Supplementary Table S5

Fig. 4

Comparison of mass spectrometry proteomic and LLFS transcriptomic results. Comparison of log2 fold changes from proteomic studies and LLFS gene expression models features characterized by either proteomic conservation analysis or transcriptomic analyses that are mappable across platforms. Proteomic results reflect differences between either offspring (left) or controls (right) to centenarian cohorts. LLFS transcriptomic results reflect age as a continuous variable, i.e., 1-year age differences. Proteins/genes annotated to the neutrophil degranulation pathway are highlighted in bold. A Comparison of the labelled LC-MS/MS study and LLFS gene expression features. The plots comprise 123 coinciding protein:gene pairs. The full set of results for these features is reported in Supplemental Table S6. B Comparison of the published SomaScan signature and LLFS gene expression features. The plots comprise 1142 coinciding protein:gene pairs. The full set of results for these features is reported in Supplemental Table S7