13 Oct 2023

GWAS uncover associations between genetic variants and human disease, but they do not reveal how those variants act on the body. Linking genetic polymorphisms to changes in plasma proteins could provide some clues. Three papers in the October 4 Nature reported one of the largest such efforts to date. The Plasma Proteomics Project, spearheaded by a consortium of 13 pharmaceutical companies, mined data from more than 50,000 participants in the U.K. Biobank to create an atlas of genome-proteome connections in a mostly European population. These data, which are available online, could point toward new drug targets and biomarkers, the authors noted.

In the first paper, scientists led by Christopher Whelan at Biogen, Melissa Miller at Pfizer, Bradford Gibson at Amgen, and Joseph Szustakowski at Bristol Myers Squibb presented data from 54,219 participants in the U.K. Biobank, a longitudinal study of people in their 40s, 50s, and 60s. These volunteers undergo genotyping, imaging, and physical exams, and provide blood and urine samples. The authors used a high-throughput commercial assay, Olink Explore 3072, to quantify nearly 3,000 plasma proteins.

First author Benjamin Sun at Biogen then correlated changes in protein level with common genetic variants. This turned up more than 14,000 protein quantitative trait loci, i.e., variants that drive protein levels up or down. Of these, 80 percent were unknown. The dataset offers insight into numerous biological processes, for example tying ABO blood type to expression of gastrointestinal proteins, and linking COVID-19 susceptibility to specific factors such as Surfactant Protein D, an immune gene expressed in the lungs.

In the second paper, AstraZeneca researchers led by Slavé Petrovski used nearly the same plasma proteome dataset, but focused on rare coding variants found by exome sequencing. Joint first authors Ryan Dhindsa and Oliver Burren at AstraZeneca, along with Sun, identified 1,962 gene-protein associations in 49,736 U.K. Biobank participants. As in the study of common variants, 80 percent were new. About a third caused truncations that suppressed the protein in question. Some associations were indirect, acting through other proteins. For example, rare coding variants in the inflammasome activator NLRC4 affected levels of pro-inflammatory cytokine IL-18, while the scavenger receptors STAB1 and STAB2 influenced levels of dozens of other proteins.

Both these studies used the Olink Explore technology, which deploys antibodies to detect proteins. However, many proteome studies use aptamers, short single-stranded oligonucleotides that fold into distinct shapes that bind specific proteins. To compare the two methods, researchers led by Kári Stefánsson and Patrick Sulem at deCODE Genetics, Reykjavik, Iceland, analyzed the plasma proteome of 35,559 Icelandic people using aptamers. About 1,500 of these participants were also in the U.K. Biobank.

Joint first authors Grimur Hjorleifsson Eldjarn and Egil Ferkingstad reported that the two approaches gave quite different results, with a correlation of only 0.33 for proteins measured in the same individuals. For example, although plasma neurofilament light protein was the top hit to associate with Alzheimer’s disease in both the aptamer and Olink data sets, the direction of the association was opposite. As measured by aptamers, NfL associated with protection; as measured with antibodies, it associated with risk. Moreover, there was no correlation between the amount of NfL measured by each methodology.

Likely, antibodies and aptamers measure different forms of the proteins, the authors concluded. These could be created by alternative splicing, post-translational modifications, cleavage, or oligomerization. A previous study also found that antibodies and aptamers detected distinct protein isoforms, often having different associations with disease (Pietzner et al., 2021). Thus, data from the two methods should be considered complementary rather than equivalent, and could help researchers home in on the precise isoforms responsible for disease.

One limitation of these studies is the largely European population in both the Icelandic and U.K. Biobank databases. In the latter, an analysis of 931 people of African ancestry, 920 Central or South Asian, 262 East Asian, and 308 Middle Eastern, turned up several disease genes that were nearly absent from the European group. For example, a truncation of the adaptive immune gene CD1C was only found in Central/South Asians, while a polymorphism that boosted levels of the insulin resistance gene SERPINA12 was common to all non-Europeans. The African population showed the most differences, sharing only two-thirds of the pQTLs found in Europeans. The findings indicate the need for larger studies of diverse populations.—Madolyn Bowman Rogers

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

Paper Citations

1. Pietzner M, Wheeler E, Carrasco-Zanini J, Kerrison ND, Oerton E, Koprulu M, Luan J, Hingorani AD, Williams SA, Wareham NJ, Langenberg C. Synergistic insights into human health from aptamer- and antibody-based proteomic profiling. Nat Commun. 2021 Nov 24;12(1):6822. PubMed.

External Citations

1. Olink Explore 3072

Further Reading

News

1. Proteins in Biofluids Foreshadow Dementia by 30 Years 29 Aug 2023
2. Proteomics Discerns Sporadic from Familial Alzheimer’s Disease 28 Jul 2023
3. Proteomics Highlight Alzheimer’s Changes in Matrisome, MAPK Signaling 3 Feb 2022
4. Massive Proteomics Study Connects Genes, Proteins, Disease 3 Dec 2021