NEW YORK – Using plasma proteomic profiles generated for UK Biobank (UKB) participants by members of a biopharmaceutical consortium known as the Pharma Proteomics Project, members of three research teams have teased out common and rare genetic variants influencing protein features in the blood, while exploring the interplay between plasma proteomics, genetics, human traits, and diseases.
"[W]e present findings from one of the largest proteomic studies conducted to date —constructing an updated genetic atlas of plasma proteome, revealing biological insights into prevalent illnesses and providing the scientific community with an open-access, population-scale proteomics resource," first and co-corresponding author Benjamin Sun, a translational sciences researcher at Biogen, and his colleagues wrote in one of the studies, published in Nature on Wednesday.
For that analysis, Sun and colleagues at Biogen, Janssen, and other research centers analyzed blood plasma measurements for 2,923 proteins in 54,219 UKB participants with directly genotyped and imputed genotyping profiles spanning some 16.1 million SNPs, unearthing almost 14,300 genetic associations involving 3,769 distinct genomic loci.
Most protein quantitative trait loci (pQTL) in the resulting collection had not been reported in the past, the team reported, noting that the diverse UKB participant group also made it possible to pick up on some ancestry-specific pQTL patterns.
"These results significantly enhance our understanding of genetic and non-genetic regulators of the proteins, uncovering many small effects at a range of additional genes, some of which play a role as interactors and components of a network regulating protein levels," Sun explained in an email, adding that the work "provides the field with both refined genetic signals and instruments for downstream investigations across diseases and hopefully seed potential experiments to further uncover the trans regulatory effects."
For another study in Nature, an international research team led by investigators at AstraZeneca searched for rare genetic variants influencing plasma protein levels using data for 49,736 UK Biobank participants with available exome sequences and blood plasma measurements for 2,923 proteins.
"The availability of exome sequences and plasma protein measurements from 50,000 UKB participants offered a unique exploration of variants across the allele frequency spectrum at unprecedented scale," co-first author Ryan Dhindsa, a researcher with AstraZeneca BioPharmaceuticals R&D's Centre for Genomics Research, said in an email.
With an exome-wide association analysis, the researchers narrowed in on 5,433 rare variant associations with protein levels, while their gene-based analysis led to more than 1,900 genes with ties to the plasma proteins. Combining these data provided still further clues to these interactions, they explained, noting that more than 99 percent of genes linked to protein levels via protein-truncating variants led to lower-than-usual levels of the proteins in question.
"We and others have previously used UK Biobank genomic data to demonstrate the importance of rare variants in common disease," co-senior author Slavé Petrovski, with AstraZeneca's Centre for Genomics Research and the University of Melbourne, said in an email.
"What's exciting about this research is that we are now able to link these high-impact rare genetic variants to effects on the human plasma proteome and in doing so accelerate our understanding of disease mechanisms, discovering biomarkers, and uncovering potential new therapeutic avenues."
In another related Nature study, investigators at Amgen's Decode Genetics and the University of Iceland considered data for roughly 50,000 UKB participants of European, African, or Asian ancestry, comparing the pQTLs identified using more than 2,900 Olink Explore 3072 platform immunoassays with those found in a prior study of Icelandic individuals profiled with more than 4,900 aptamer-based assays encompassed in the SomaScan v4 platform.
"Our results show that the value of the information provided by the two most commonly used high-throughput proteomics platforms and demonstrate the differences between them that at times provides useful complementarity," Decode Genetics researchers Kári Stefánsson, Patrick Sulem, and Grímur Hjörleifsson Eldjárn, the study's co-senior authors and co-first author, said in an email.
Though the Olink and SomaScan platforms unearthed a similar number of cis-acting pQTLs mediating the activity of nearby proteins, for example, the team found that more of the Olink assays showed evidence for the pQTLs. In a set of 1,848 proteins profiled with both platforms, pQTL evidence was unearthed with 80 percent of the assays in Olink and 58 percent of SomaScan assays.
"While these two proteomics platforms serve as useful instruments for simultaneous testing of thousands of proteins in large datasets, the discrepancies between platforms can affect the conclusions drawn in the context of diseases," Stefánsson, Sulem, and Hjörleifsson Eldjárn said. "Therefore, careful validation is necessary for individual proteins on a case-by-case basis."
The platform effects may be particularly pertinent when attempting to assess proteins found at relatively low levels in blood plasma, the investigators explained, noting that the platform selected also impacted the genetic variants linked to plasma protein levels in some cases.
They emphasized the potential of digging into the diverse genetic contributors to plasma protein levels to better understand human traits and diseases, while better teasing out protein level diversity patterns that coincide with environmental rather than genetic factors.
"When the level of a protein correlates with a disease it can either be a consequence of the disease or the protein may participate in the pathogenesis of the disease," Stefánsson, Sulem, and Hjörleifsson Eldjárn noted. "One way to separate the two possibilities is to determine if the variants in the sequence that affect the level of the protein also correlate with the disease."
"In both instances the protein could serve as a biomarker and if the variants correlate with the disease, it could also serve as a drug target," they added, noting that the "incredible wealth of phenotypic data in the UKB makes these proteomic data useful in the analysis of a very large number of diseases and physiologic traits."