Single-cell map links rare and common DNA variants to immune gene activity

Two people can carry the same genes yet use them to different degrees, and part of that difference is written in their DNA. A study published in Nature charts this inherited variation in gene activity across the immune system one cell at a time. It draws on more than five million cells, spanning 28 immune cell types, from 1,925 individuals. According to the authors, common genetic variants appear to influence the activity of most of the genes they examined, and rarer variants leave a detectable trace on almost half. Roughly half of the effects show up in only one or a few cell types.

The work is the first phase of a project called TenK10K, reported by Anna Cuomo and colleagues at institutions including the Garvan Institute of Medical Research, the Murdoch Children's Research Institute, the University of Sydney and UNSW Sydney. It combines two kinds of measurement taken from the same people. Whole-genome sequencing reads each person's full DNA sequence and reveals the variants, or spelling differences, they carry. Single-cell RNA sequencing captures the RNA molecules inside individual cells. That shows which genes are switched on and how strongly, which geneticists call gene expression. Linking the two lets researchers find expression quantitative trait loci, or eQTLs. These are positions in the genome where differences between people track differences in how much a gene is expressed.

Most such mapping has focused on common variants, which are shared by a sizeable fraction of the population. Rare and low-frequency variants are carried by few people, so their effects are hard to detect without large samples. The authors note that their impact on cell-specific gene expression has largely gone uncharacterised. Scale helps, and so does method. Single-cell studies have often added up the counts from all cells of a given type in each person, producing so-called pseudobulk data. The team instead used a recent method that models single cells directly, which, the authors say, increases the power to detect associations.

The authors tested 21,404 genes. For 83% of them they identify putative common regulatory variants, meaning candidate variants that appear to adjust a gene's activity. They also detect cumulative rare-variant signals, which assess the combined contribution of rare variants, for 47% of those genes. To ask how far these effects are confined to particular cells, they developed a framework for measuring how cell-type-specific each eQTL is. About half turn out to be seen in only one or a few cell types. The team also looked beyond tidy cell categories at continuous cell states, the gradual shifts a cell can pass through. They examined how genetic effects change along those states and identified variants that alter how abundant particular cell states are.

The team then combined its results with disease data and with functional annotations, which is existing information about what different stretches of the genome do. This let them characterise likely molecular modes of action for many links between variants and disease. These are plausible accounts of how a disease-linked variant might exert its influence by changing a gene's expression in specific immune cells.

The abstract describes these as putative variants and likely mechanisms. They are leads that would need further work to confirm, not established causes. The abstract does not say which diseases are involved or how individual findings were validated. As the first phase of TenK10K, the study also presents itself as the start of a larger effort, and later phases may add to these results.