Every person carries two copies of most genes, with one version, or allele, coming from each parent. Even though these copies tend to be functionally redundant, conventional genetic theory dictates that biallelic expression, wherein both alleles are transcribed equally when the gene is expressed, is nearly universal. But a study published in Cell Reports in January posits something different: that it might be somewhat common for cells to preferentially express only one of a gene’s alleles.
Using a model that characterizes patterns in allele expression from bulk RNA sequencing data, the researchers found that both alleles of some genes were equally active throughout tissues—but for almost 3,000 genes, cells favored one allele or the other, exhibiting what the authors call random allelic expression (RAE). These genes are more likely to acquire harmful mutations, the authors say, and therefore, could act as biomarkers for predicting or diagnosing disease.
Brian Chen, a biologist at McGill University, says the study was “very interesting,” adding: “I definitely thought that the approach was really clever.” Although Chen wasn’t alone in praising the work, the findings sparked some controversy.
How common is genetic favoritism?
Biased allele expression is a known genetic phenomenon, but it’s thought to be rare. There are some more common exceptions, however, such as how cells with two X chromosomes inactivate one of them during development in order to meter the expression level of X-linked genes that can be toxic in double doses. And scientists have described cases of random monoallelic expression, wherein cells only express one allele and suppress the other, but experts disagree about the phenomenon’s prevalence. Some studies have suggested it occurs in as many as 10 percent of genes, however, Rickard Sandberg, a geneticist at the Karolinska Institute in Sweden, says that his experiments have suggested that imbalanced expression likely occurs in less than one percent of genes.
Christopher Gregg, a neurobiologist and geneticist at the University of Utah school of medicine, suspected that biased allele activity is more common than the literature suggests. His lab previously found widespread evidence of allelic imbalances in the brains of mice and in cultured cells. “[We] discovered that there are some genes where the two alleles are absolutely in lockstep with each other,” Gregg explains. “But there were other genes that were completely decoupled.”
Wanting to extend the findings to human tissue, he enlisted his former postdoc Stephanie Kravitz to analyze a bulk RNAseq dataset from the genotype tissue expression (GTEx) consortium that included sequences from more than 15,000 genes in 54 tissues from 832 human donors. From this data, Kravitz calculated the abundance of each allele across tissues.
The researchers then compared the relative gene expression levels of each allele across different tissues within each individual, theorizing that if two alleles are expressed in a biallelic fashion, the overall allelic expression ratio should fit a binomial distribution, whereas randomly expressed alleles wouldn’t. As a control, they tested whether the method could distinguish X-linked genes from autosomal genes, and indeed, the model labeled X-linked genes as RAE, validating the model’s ability to discern biased expression of alleles.
In all, the analysis picked out 2,762 non–X-linked genes that fit the RAE pattern. If accurate, that number represents almost 10 percent of known human genes. By performing a gene ontology analysis, which annotates the molecular targets and biological processes associated with a group of genes, the researchers found that RAE genes were more likely to be involved in immune adaptability and cellular plasticity than biallelic genes. Biallelic genes, on the other hand, are much more evolutionarily conserved and are more likely to be associated with cell survival, Gregg explains, meaning they’re less likely to tolerate mutations. The researchers also found that biallelic genes tend to be located near the centromere, indicating that they’re less likely to undergo recombination, while random allelic genes are more likely to be clustered near the telomere’s tips.
The ontology analysis also revealed that RAE genes are associated with disease, Gregg says. “They are preferentially enriched around genes that are [associated with] age-related diseases like cardiovascular disease, cancer, and heart disease,” and therefore, understanding where and why one allele gets expressed more than its counterpart could lead to more accurate health risk assessments or even diagnoses, he posits.
Moreover, Gregg says, the findings challenge the idea that all of a person’s cells work with the same set of genetic blueprints. “If it’s the case that there are some cells that are expressing this allele and other cells that are expressing that allele, then they actually have different genotypes,” he explains. “That’s really kind of mind-blowing.”
Chen says that, as an experimentalist, it “makes sense” to him that some genes would be more tightly regulated than others. He adds that studies like this may help scientists narrow in on potential disease-causing genes, which then could become targets for future study.
Alexander Mendenhall, a geneticist at the University of Washington, writes in an email to The Scientist that the study is “wonderous” and that the authors have “a good positive control with X inactivation,” which is an exciting result. He agrees with Gregg that the study has translational potential. “It suggests that part of a patient’s personalized medicine profile may need to be their gene expression profile, at a level that incorporates these” allelic biases.
However, not everyone is convinced. Sandberg points out that there is, as of yet, no mechanistic basis behind what the authors call random allelic expression. “It’s not founded on well-established concepts, such as imprinting, X-chromosome inactivation, allelic imbalance, or random monoallelic expression,” he notes, and therefore, he questions whether the findings are meaningful. “We don’t know if it’s even appropriate to use” the mathematical model the authors used in the paper, as he speculates it may pick up on unrelated expression phenomena that could be explained by cell to cell variation, such as cell-specific alternative splicing. He says other methods, such as single-cell RNAseq, could give researchers more information about what cell types they’re looking at and therefore help distinguish among phenomena.
Gregg admits that the mechanisms behind random allelic expression are unknown, but stands by the methodology the team employed. He argues that single-cell methods are useful, but have “limited applications” to profile tissues throughout the body, and can yield “technically noisy data.” He adds because RAE genes are more tolerant to mutations and are associated with disease, they show “important new links,” between RAE and the factors that drive the differences between human beings and are targets for future studies.
Gregg also argues that there are likely multiple mechanisms at play. For instance, he points to transcriptional interference, wherein adjacent genes with overlapping sequences can’t be active simultaneously. Another possibility is enhancer interference, wherein two genes compete to use the same enhancer. “The truth is that we don’t know the cause” of RAE, says Gregg. “There likely won’t be one solution that explains the whole thing.”
In addition to ferreting out the mechanisms of RAE, Gregg says he hopes to continue studying the phenomenon in human tissues and its connection with diseases such as cancer. He theorizes that RAE could be a biomarker for how dangerous tumors are or how well the body copies DNA as we age.