bioRxiv Science⌕ Search

Biology subjects

Liu, T. A.

Publications and source records attributed to Liu, T. A..

3 recordsLinked to original sources

Enhancer Poising Enables Pathogenic Gene Activation by Noncoding Variants

Single nucleotide variants within enhancers--non-coding DNA elements that regulate transcription--often lead to aberrant gene activation and contribute to a wide range of genetic disorders (Claussnitzer et al. 2015; Doan et al. 2016; Turner et al. 2017; Yanchus et al. 2022; Lettice et al. 2008). The mechanism by which ectopic gene activation occurs through these gain-of-function enhancer mutations remains poorly understood. Using the ZRS, a benchmark disease-associated enhancer of Sonic hedgehog (Shh), as a model, we demonstrate that poised (i.e., accessible but inactive) chromatin sensitizes Shh to aberrant activation in anterior limb bud, leading to polydactyly. In the anterior limb cells of wild-type mice, Shh is inactive, but the ZRS is accessible and marked by enhancer-associated histone modifications. We demonstrate that this poising signature explains how over 20 independent rare variants within the ZRS cause Shh misexpression in the same anterior limb bud cell population, resulting in similar limb malformations, despite affecting binding sites for different activators and repressors. Disabling pioneer transcription factor binding to the ZRS suppresses its poised state in anterior cells, prevents aberrant activation of the ZRS by rare variants, and fully rescues limb malformations in variant knock-in mice. A thorough examination of other disease-associated enhancers with pathogenic gain-of-function variants revealed that they are all poised in tissues with ectopic activity. We use this poising signature to predict and validate in vivo ectopic forebrain activity of previously uncharacterized autism-associated non-coding variants. Our findings suggest that spatial enhancer poising, likely a byproduct of development, creates a susceptibility to non-coding mutations and offers a potential mechanistic explanation for the burden of disease-associated non-coding variants.

genomics↗

Rapid and Quantitative Functional Interrogation of Human Enhancer Variant Activity in Live Mice

Functional analysis of non-coding variants associated with human congenital disorders remains challenging due to the lack of efficient in vivo models. Here we introduce dual-enSERT, a robust Cas9-based two-color fluorescent reporter system which enables rapid, quantitative comparison of enhancer allele activities in live mice of any genetic background. We use this new technology to examine and measure the gain- and loss-of-function effects of enhancer variants linked to limb polydactyly, autism, and craniofacial malformation. By combining dual-enSERT with single-cell transcriptomics, we characterize variant enhancer alleles at cellular resolution, thereby implicating candidate molecular pathways in pathogenic enhancer misregulation. We further show that independent, polydactyly-linked enhancer variants lead to ectopic expression in the same cell populations, indicating shared genetic mechanisms underlying non-coding variant pathogenesis. Finally, we streamline dual-enSERT for analysis in F0 animals by placing both reporters on the same transgene separated by a synthetic insulator. Dual-enSERT allows researchers to go from identifying candidate enhancer variants to analysis of comparative enhancer activity in live embryos in under two weeks.

genomics↗

Partitioning gene-based variance of complex traits by gene score regression

The majority of genome-wide association studies (GWAS) loci are not annotated to known genes in the human genome, which renders biological interpretations difficult. Transcriptome-wide association studies (TWAS) associate complex traits with genotype-based prediction of gene expression deriving from expression quantitative loci(eQTL) studies, thus improving the interpretability of GWAS findings. However, these results can sometimes suffer from a high false positive rate, because predicted expression of different genes may be highly correlated due to linkage disequilibrium between eQTL. We propose a novel statistical method, Gene Score Regression (GSR), to detect causal gene sets for complex traits while accounting for gene-to-gene correlations. We consider non-causal genes that are highly correlated with the causal genes will also exhibit a high marginal association with the complex trait. Consequently, by regressing on the marginal associations of complex traits with the sum of the gene-to-gene correlations in each gene set, we can assess the amount of variance of the complex traits explained by the predicted expression of the genes in each gene set and identify plausible causal gene sets. GSR can operate either on GWAS summary statistics or observed gene expression. Therefore, it may be widely applied to annotate GWAS results and identify the underlying biological pathways. We demonstrate the high accuracy and computational efficiency of GSR compared to state-of-the-art methods through simulations and real data applications. GSR is openly available at https://github.com/li-lab-mcgill/GSR.

bioinformatics↗