Two Wrong Genetic Variants Can Restore Health and Challenge Long-Held Disease Assumptions
Scientists at the Pacific Northwest Research Institute (PNRI) have uncovered a surprising and important insight into human genetics: inheriting two damaging variants of the same gene does not always worsen disease. In many cases, those two harmful variants can actually restore normal protein function, directly challenging one of the most basic assumptions used in genetic research and clinical diagnosis.
This discovery reshapes how scientists think about genetic mutations, disease risk, and the way genetic tests are interpreted—especially in the context of rare inherited disorders.
Rethinking a Core Assumption in Genetics
For decades, genetic science has largely operated on a straightforward idea: if one harmful variant of a gene causes disease, then having two harmful variants should make things even worse. This logic underpins many diagnostic tools and risk prediction models used in medicine today.
However, the PNRI research team found that this assumption does not always hold true. Their work shows that genetic variants do not always act independently. Instead, interactions between variants can fundamentally change biological outcomes.
The study demonstrates that, in some cases, two variants that are damaging on their own can work together in a way that restores healthy protein activity.
The Enzyme at the Center of the Study
The research focused on a human enzyme called argininosuccinate lyase (ASL). This enzyme plays a critical role in the urea cycle, a vital metabolic pathway responsible for removing toxic ammonia from the body.
When ASL does not function properly, ammonia can build up in the bloodstream, leading to urea cycle disorders. These are rare but potentially life-threatening conditions that can cause neurological damage, organ failure, or even death if untreated.
Variants that reduce ASL activity are a known cause of these disorders, making ASL an ideal candidate for studying how genetic mutations affect protein function.
Thousands of Variants, One Surprising Pattern
Instead of examining just a handful of mutations, the PNRI team took a large-scale experimental approach. They measured the functional impact of several thousand individual ASL variants, as well as thousands of variant combinations.
The results were striking. More than 60% of variant pairs that were individually damaging showed restored enzyme activity when present together. In many cases, the combined activity approached levels seen in healthy individuals.
This means that a person carrying two different harmful variants of ASL might not experience the severe disease outcomes traditionally expected.
Intragenic Complementation Comes Into Focus
The phenomenon behind this effect is known as intragenic complementation. It occurs when two variants in different parts of the same protein compensate for each other’s weaknesses.
The idea itself is not new. It was first proposed in 1964 by Francis Crick and Leslie Orgel, two influential figures in molecular biology. However, until now, intragenic complementation had never been tested systematically across thousands of variants, nor shown to be common and predictable in human genes.
This study provides the strongest evidence yet that intragenic complementation is not a rare curiosity—it is a biologically meaningful and widespread mechanism.
Why Structure Matters
Proteins like ASL are not simple chains. They fold into complex three-dimensional structures, and their function often depends on how different regions interact with one another.
The PNRI researchers found that complementation tends to occur when two variants affect different structural or functional regions of the protein. One variant may disrupt one part of the protein, while the second variant disrupts another part—but together, the overall structure regains enough stability to function properly.
This structural logic explains why two “wrong” changes can sometimes make things right again.
Artificial Intelligence Makes Prediction Possible
Understanding these interactions is one thing. Predicting them is another.
To address this, the team developed an AI-based predictive model designed to forecast whether pairs of variants would restore protein function. The model uses structural and functional information about proteins to make its predictions.
The results were remarkably strong. The AI model achieved nearly 100% accuracy in predicting intragenic complementation in ASL. To test whether the rules extended beyond a single enzyme, the researchers applied the model to a second human enzyme, fumarase.
Once again, the predictions held up, suggesting that the same principles apply across different proteins.
Implications for Genetic Testing and Medicine
One of the most important outcomes of this work lies in how genetic information is interpreted in clinical settings.
Current genetic testing often evaluates variants one at a time, classifying them as benign, harmful, or of uncertain significance. This study shows that such an approach can overestimate disease risk, particularly for individuals who carry two different variants in the same gene.
The researchers estimate that about 4% of human genes have the structural features needed for intragenic complementation. For these genes, standard prediction models may need serious revision.
This has direct implications for:
- Rare disease diagnosis
- Carrier screening
- Genetic counseling
- Personalized medicine
In some cases, patients previously thought to be at high risk may actually have functional proteins, thanks to variant interactions that current tools fail to recognize.
A Collaborative Effort Across Institutions
The study was led by scientists in the Dudley Lab at PNRI, with collaboration from multiple institutions, including:
- Children’s National Hospital
- St. Jude Children’s Research Hospital
- George Mason University
- University of Washington
This interdisciplinary effort combined molecular biology, computational modeling, clinical expertise, and artificial intelligence.
Understanding Epistasis in Human Genetics
At a broader level, this research contributes to the study of epistasis, which refers to interactions between genetic variants that alter their combined effects.
While epistasis has long been recognized in theory, it has been difficult to measure and predict in real biological systems. This work shows that epistatic interactions are not only real, but structured, predictable, and clinically relevant.
Why This Discovery Matters Going Forward
This research challenges the idea that genetic damage always adds up in a simple way. Instead, it highlights the complex, interactive nature of biology, where context and structure matter just as much as individual mutations.
As genetic sequencing becomes more common worldwide, understanding how variants interact will be essential for avoiding misdiagnosis and improving patient care.
The takeaway is clear: in genetics, two wrongs can sometimes make a right.
Research Paper:
https://www.pnas.org/doi/10.1073/pnas.2516291123
