Allele-specific protein complex assembly counteracts the dominant negative effect

Homomers are protein complexes made up of multiple copies (subunits) of the same type of protein. For a long time, it has been theorised that cotranslational assembly may influence the subunit composition of homomeric complexes if both wild-type and mutant alleles are present.

When cotranslational assembly occurs, two or more subunits of the complex bind each other during their translation on the ribosome. For homomers, this process is thought to happen on the same mRNA, meaning that the resultant complex will either be purely wild type or mutant.

Mihaly wanted to find out if this mechanism is able to "buffer" the effect of dominant-negative (DN) mutations. These mutations transfer their functionally damaging effect to the wild type as the mutant subunits co-assemble with wild-type subunits into the same complex.

To explore this, they looked at the inheritance of mutations in homomers. Because DN mutations are dominantly inherited, they expected that homomers with dominant (AD) disease mutations should be depleted in cotranslationally assembly relative to recessive inheritance (AR).

They found that AD homomers are much less likely to cotranslationally assemble. The trend holds up to a range of potential confounders, such as cellular abundance, protein complex symmetry, and hemizygosity.

To understand how the underlying molecular disease mechanisms are affected, the team classified the majority of known AD genes into one of the protein-level disease mechanisms: loss/gain-of-function (LOF, GOF) and DN.

With this dataset, they observed a statistically significant depletion of cotranslational assembly in homomers whose mutations predominantly exert a DN effect, relative to those homomers whose mutations are thought to act via a simple LOF.

Interestingly, they found that homodimers with DN mutations tend to have C-terminally exposed interfaces, increasing the chance of the assembly happening only after translation of the subunits is complete.

Previous work by the Joe Marsh Lab has shown that non-LOF (e.g., GOF and DN) mutations are consistently underpredicted by variant effect predictors, highlighting the importance of considering alternative mechanisms in the effort to annotate the pathogenic variation.

The team built a computational model to predict proteins whose mutations are more likely to be non-LOF than simple heterozygous LOF. They provide predictions covering nearly half of the human proteome and it is hoped this will prove a valuable resource to clinical geneticists.

Related Links

• Buffering of genetic dominance by allele-specific protein complex assembly published in Science Advances: 10.1126/sciadv.adf9845
• Joe Marsh Research Group