Summary Distant-acting tissue-specific enhancers vastly outnumber protein-coding genes in mammalian genomes, but the functional significance of this regulatory complexity remains insufficiently understood1,2. Here we show that the pervasive presence of multiple enhancers with similar activities near the same gene confers phenotypic robustness to loss-of-function mutations in individual enhancers. We used genome editing to create 23 mouse deletion lines and inter-crosses, including both single and combinatorial enhancer deletions at seven distinct loci required for limb development. Surprisingly, none of ten deletions of individual enhancers caused noticeable changes in limb morphology. In contrast, removal of pairs of limb enhancers near the same gene resulted in discernible phenotypes, indicating that enhancers function redundantly in establishing normal morphology. In a genetic background sensitized by reduced baseline expression of the target gene, even single enhancer deletions caused limb abnormalities, suggesting that functional redundancy is conferred by additive effects of enhancers on gene expression levels. A genome-wide analysis integrating epigenomic and transcriptomic data from 29 developmental mouse tissues revealed that mammalian genes are very commonly associated with multiple enhancers that have similar spatiotemporal activity. Systematic exploration of three representative developmental structures (limb, brain, heart) uncovered more than a thousand cases in which five or more enhancers with redundant activity patterns were found near the same gene. Taken together, our data indicate that enhancer redundancy is a remarkably widespread feature of mammalian genomes and provides an effective regulatory buffer preventing deleterious phenotypic consequences upon loss of individual enhancers.

The Encyclopedia of DNA Elements (ENCODE) project has established a genomic resource for mammalian development, profiling a diverse panel of mouse tissues at 8 developmental stages from 10.5 days after conception until birth, including transcriptomes, methylomes and chromatin states. Here we systematically examined the state and accessibility of chromatin in the developing mouse fetus. In total we performed 1,128 chromatin immunoprecipitation with sequencing (ChIP–seq) assays for histone modifications and 132 assay for transposase-accessible chromatin using sequencing (ATAC–seq) assays for chromatin accessibility across 72 distinct tissue-stages. We used integrative analysis to develop a unified set of chromatin state annotations, infer the identities of dynamic enhancers and key transcriptional regulators, and characterize the relationship between chromatin state and accessibility during developmental gene regulation. We also leveraged these data to link enhancers to putative target genes and demonstrate tissue-specific enrichments of sequence variants associated with disease in humans. The mouse ENCODE data sets provide a compendium of resources for biomedical researchers and achieve, to our knowledge, the most comprehensive view of chromatin dynamics during mammalian fetal development to date.

Non-coding "ultraconserved" regions containing hundreds of consecutive bases of perfect sequence conservation across mammalian genomes can function as distant-acting enhancers. However, initial deletion studies in mice revealed that loss of such extraordinarily constrained sequences had no immediate impact on viability. Here, we show that ultraconserved enhancers are required for normal development. Focusing on some of the longest ultraconserved sites genome wide, located near the essential neuronal transcription factor Arx, we used genome editing to create an expanded series of knockout mice lacking individual or combinations of ultraconserved enhancers. Mice with single or pairwise deletions of ultraconserved enhancers were viable and fertile but in nearly all cases showed neurological or growth abnormalities, including substantial alterations of neuron populations and structural brain defects. Our results demonstrate the functional importance of ultraconserved enhancers and indicate that remarkably strong sequence conservation likely results from fitness deficits that appear subtle in a laboratory setting.

Establishing causal links between non-coding variants and human phenotypes is an increasing challenge. Here we introduce a high-throughput mouse reporter assay for assessing the pathogenic potential of human enhancer variants in vivo and examine nearly a thousand variants in an enhancer repeatedly linked to polydactyly. We show that 71% of all rare non-coding variants previously proposed as causal led to reporter gene expression in a pattern consistent with their pathogenic role. Variants observed to alter enhancer activity were further confirmed to cause polydactyly in knock-in mice. We also used combinatorial and single-nucleotide mutagenesis to evaluate the in vivo impact of mutations affecting all positions of the enhancer and identified additional functional substitutions, including potentially pathogenic variants hitherto not observed in humans. Our results uncover the functional consequences of hundreds of mutations in a phenotype-associated enhancer and establish a widely applicable strategy for systematic in vivo evaluation of human enhancer variants.

Enhancers are important noncoding elements, but they have been traditionally hard to characterize experimentally. The development of massively parallel assays allows the characterization of large numbers of enhancers for the first time. Here, we developed a framework using Drosophila STARR-seq to create shape-matching filters based on meta-profiles of epigenetic features. We integrated these features with supervised machine-learning algorithms to predict enhancers. We further demonstrated our model could be transferred to predict enhancers in mammals. We comprehensively validated the predictions using a combination of in vivo and in vitro approaches, involving transgenic assays in mouse and transduction-based reporter assays in human cell lines (153 enhancers in total). The results confirmed our model can accurately predict enhancers in different species without re-parameterization. Finally, we examined the transcription-factor binding patterns at predicted enhancers versus promoters. We demonstrated that these patterns enable the construction of a secondary model effectively discriminating between enhancers and promoters.