Genome-wide functional screen of 3'UTR variants uncovers causal variants for human disease and evolution.

3' untranslated region (3'UTR) variants are strongly associated with human traits and diseases, yet few have been causally identified. We developed the massively parallel reporter assay for 3'UTRs (MPRAu) to sensitively assay 12,173 3'UTR variants. We applied MPRAu to six human cell lines, focusing on genetic variants associated with genome-wide association studies (GWAS) and human evolutionary adaptation. MPRAu expands our understanding of 3'UTR function, suggesting that simple sequences predominately explain 3'UTR regulatory activity. We adapt MPRAu to uncover diverse molecular mechanisms at base pair resolution, including an adenylate-uridylate (AU)-rich element of LEPR linked to potential metabolic evolutionary adaptations in East Asians. We nominate hundreds of 3'UTR causal variants with genetically fine-mapped phenotype associations. Using endogenous allelic replacements, we characterize one variant that disrupts a miRNA site regulating the viral defense gene TRIM14 and one that alters PILRB abundance, nominating a causal variant underlying transcriptional changes in age-related macular degeneration.

A Deep Neural Network for Predicting and Engineering Alternative Polyadenylation.

Alternative polyadenylation (APA) is a major driver of transcriptome diversity in human cells. Here, we use deep learning to predict APA from DNA sequence alone. We trained our model (APARENT, APA REgression NeT) on isoform expression data from over 3 million APA reporters. APARENT's predictions are highly accurate when tasked with inferring APA in synthetic and human 3'UTRs. Visualizing features learned across all network layers reveals that APARENT recognizes sequence motifs known to recruit APA regulators, discovers previously unknown sequence determinants of 3' end processing, and integrates these features into a comprehensive, interpretable, cis-regulatory code. We apply APARENT to forward engineer functional polyadenylation signals with precisely defined cleavage position and isoform usage and validate predictions experimentally. Finally, we use APARENT to quantify the impact of genetic variants on APA. Our approach detects pathogenic variants in a wide range of disease contexts, expanding our understanding of the genetic origins of disease.

Systematic analysis of intrinsic enhancer-promoter compatibility in the mouse genome.

Gene expression is in part controlled by cis-regulatory elements (CREs) such as enhancers and repressive elements. Anecdotal evidence has indicated that a CRE and a promoter need to be biochemically compatible for promoter regulation to occur, but this compatibility has remained poorly characterized in mammalian cells. We used high-throughput combinatorial reporter assays to test thousands of CRE-promoter pairs from three Mb-sized genomic regions in mouse cells. This revealed that CREs vary substantially in their promoter compatibility, ranging from striking specificity to broad promiscuity. More than half of the tested CREs exhibit significant promoter selectivity. Housekeeping promoters tend to have similar CRE preferences, but other promoters exhibit a wide diversity of compatibilities. Higher-order transcription factors (TF) motif combinations may account for compatibility. CRE-promoter selectivity does not correlate with looping interactions in the native genomic context, suggesting that chromatin folding and compatibility are two orthogonal mechanisms that confer specificity to gene regulation.

Strategies for dissecting the complexity of neurodevelopmental disorders.

Neurodevelopmental disorders (NDDs) are associated with a wide range of clinical features, affecting multiple pathways involved in brain development and function. Recent advances in high-throughput sequencing have unveiled numerous genetic variants associated with NDDs, which further contribute to disease complexity and make it challenging to infer disease causation and underlying mechanisms. Herein, we review current strategies for dissecting the complexity of NDDs using model organisms, induced pluripotent stem cells, single-cell sequencing technologies, and massively parallel reporter assays. We further highlight single-cell CRISPR-based screening techniques that allow genomic investigation of cellular transcriptomes with high efficiency, accuracy, and throughput. Overall, we provide an integrated review of experimental approaches that can be applicable for investigating a broad range of complex disorders.

Systematic identification of genotype-dependent enhancer variants in eosinophilic esophagitis.

Eosinophilic esophagitis (EoE) is a rare atopic disorder associated with esophageal dysfunction, including difficulty swallowing, food impaction, and inflammation, that develops in a small subset of people with food allergies. Genome-wide association studies (GWASs) have identified 9 independent EoE risk loci reaching genome-wide significance (p < 5 × 10-8) and 27 additional loci of suggestive significance (5 × 10-8 < p < 1 × 10-5). In the current study, we perform linkage disequilibrium (LD) expansion of these loci to nominate a set of 531 variants that are potentially causal. To systematically interrogate the gene regulatory activity of these variants, we designed a massively parallel reporter assay (MPRA) containing the alleles of each variant within their genomic sequence context cloned into a GFP reporter library. Analysis of reporter gene expression in TE-7, HaCaT, and Jurkat cells revealed cell-type-specific gene regulation. We identify 32 allelic enhancer variants, representing 6 genome-wide significant EoE loci and 7 suggestive EoE loci, that regulate reporter gene expression in a genotype-dependent manner in at least one cellular context. By annotating these variants with expression quantitative trait loci (eQTL) and chromatin looping data in related tissues and cell types, we identify putative target genes affected by genetic variation in individuals with EoE. Transcription factor enrichment analyses reveal possible roles for cell-type-specific regulators, including GATA3. Our approach reduces the large set of EoE-associated variants to a set of 32 with allelic regulatory activity, providing functional insights into the effects of genetic variation in this disease.

Functional characterization of human genomic variation linked to polygenic diseases.

The burden of human disease lies predominantly in polygenic diseases. Since the early 2000s, genome-wide association studies (GWAS) have identified genetic variants and loci associated with complex traits. These have ranged from variants in coding sequences to mutations in regulatory regions, such as promoters and enhancers, as well as mutations affecting mediators of mRNA stability and other downstream regulators, such as 5' and 3'-untranslated regions (UTRs), long noncoding RNA (lncRNA), and miRNA. Recent research advances in genetics have utilized a combination of computational techniques, high-throughput in vitro and in vivo screening modalities, and precise genome editing to impute the function of diverse classes of genetic variants identified through GWAS. In this review, we highlight the vastness of genomic variants associated with polygenic disease risk and address recent advances in how genetic tools can be used to functionally characterize them.

Massively parallel reporter assays and variant scoring identified functional variants and target genes for melanoma loci and highlighted cell-type specificity.

The most recent genome-wide association study (GWAS) of cutaneous melanoma identified 54 risk-associated loci, but functional variants and their target genes for most have not been established. Here, we performed massively parallel reporter assays (MPRAs) by using malignant melanoma and normal melanocyte cells and further integrated multi-layer annotation to systematically prioritize functional variants and susceptibility genes from these GWAS loci. Of 1,992 risk-associated variants tested in MPRAs, we identified 285 from 42 loci (78% of the known loci) displaying significant allelic transcriptional activities in either cell type (FDR < 1%). We further characterized MPRA-significant variants by motif prediction, epigenomic annotation, and statistical/functional fine-mapping to create integrative variant scores, which prioritized one to six plausible candidate variants per locus for the 42 loci and nominated a single variant for 43% of these loci. Overlaying the MPRA-significant variants with genome-wide significant expression or methylation quantitative trait loci (eQTLs or meQTLs, respectively) from melanocytes or melanomas identified candidate susceptibility genes for 60% of variants (172 of 285 variants). CRISPRi of top-scoring variants validated their cis-regulatory effect on the eQTL target genes, MAFF (22q13.1) and GPRC5A (12p13.1). Finally, we identified 36 melanoma-specific and 45 melanocyte-specific MPRA-significant variants, a subset of which are linked to cell-type-specific target genes. Analyses of transcription factor availability in MPRA datasets and variant-transcription-factor interaction in eQTL datasets highlighted the roles of transcription factors in cell-type-specific variant functionality. In conclusion, MPRAs along with variant scoring effectively prioritized plausible candidates for most melanoma GWAS loci and highlighted cellular contexts where the susceptibility variants are functional.

Proformer: a hybrid macaron transformer model predicts expression values from promoter sequences.

The breakthrough high-throughput measurement of the cis-regulatory activity of millions of randomly generated promoters provides an unprecedented opportunity to systematically decode the cis-regulatory logic that determines the expression values. We developed an end-to-end transformer encoder architecture named Proformer to predict the expression values from DNA sequences. Proformer used a Macaron-like Transformer encoder architecture, where two half-step feed forward (FFN) layers were placed at the beginning and the end of each encoder block, and a separable 1D convolution layer was inserted after the first FFN layer and in front of the multi-head attention layer. The sliding k-mers from one-hot encoded sequences were mapped onto a continuous embedding, combined with the learned positional embedding and strand embedding (forward strand vs. reverse complemented strand) as the sequence input. Moreover, Proformer introduced multiple expression heads with mask filling to prevent the transformer models from collapsing when training on relatively small amount of data. We empirically determined that this design had significantly better performance than the conventional design such as using the global pooling layer as the output layer for the regression task. These analyses support the notion that Proformer provides a novel method of learning and enhances our understanding of how cis-regulatory sequences determine the expression values.

Using Synthetic DNA Libraries to Investigate Chromatin and Gene Regulation.

Despite the recent explosion in genome-wide studies in chromatin and gene regulation, we are still far from extracting a set of genetic rules that can predict the function of the regulatory genome. One major reason for this deficiency is that gene regulation is a multi-layered process that involves an enormous variable space, which cannot be fully explored using native genomes. This problem can be partially solved by introducing synthetic DNA libraries into cells, a method that can test the regulatory roles of thousands to millions of sequences with limited variables. Here, we review recent applications of this method to study transcription factor (TF) binding, nucleosome positioning, and transcriptional activity. We discuss the design principles, experimental procedures, and major findings from these studies and compare the pros and cons of different approaches.

Altered regulation of DPF3, a member of the SWI/SNF complexes, underlies the 14q24 renal cancer susceptibility locus.

Our study investigated the underlying mechanism for the 14q24 renal cell carcinoma (RCC) susceptibility risk locus identified by a genome-wide association study (GWAS). The sentinel single-nucleotide polymorphism (SNP), rs4903064, at 14q24 confers an allele-specific effect on expression of the double PHD fingers 3 (DPF3) of the BAF SWI/SNF complex as assessed by massively parallel reporter assay, confirmatory luciferase assays, and eQTL analyses. Overexpression of DPF3 in renal cell lines increases growth rates and alters chromatin accessibility and gene expression, leading to inhibition of apoptosis and activation of oncogenic pathways. siRNA interference of multiple DPF3-deregulated genes reduces growth. Our results indicate that germline variation in DPF3, a component of the BAF complex, part of the SWI/SNF complexes, can lead to reduced apoptosis and activation of the STAT3 pathway, both critical in RCC carcinogenesis. In addition, we show that altered DPF3 expression in the 14q24 RCC locus could influence the effectiveness of immunotherapy treatment for RCC by regulating tumor cytokine secretion and immune cell activation.

The Search for and Functional Analysis of Genetic Variants in microRNA-Binding Sites using Massively Parallel Reporter Assay.

A large-scale search for the genetic variants with a bias in the representation of alleles in transcriptome data (AE SNPs) and the binding sites in microRNA 3'-UTRs was performed and their functional significance was assessed using massively parallel reporter assay (MPRA). Of the 629,559 associated "SNP-gene" pairs (eQTLs) discovered in the human liver tissue according to the GTEx Analysis V8 data, 4394 polymorphic positions in the 3'-UTRs of the genes, which represent the eQTLs for these genes were selected. The TargetScanHuman 7.0 algorithm and PolymiRTS database were searched for the potential microRNA-binding sites. Of the predicted microRNA sites affected by eQTL-SNPs, we selected 51 sites with the best evidence of functionality according to Ago2-CLIP-seq, CLEAR-CLIP, and eCLIP-seq for RNA-binding proteins. For MPRA, a library of the plasmids carrying the main and alternative alleles for each AE SNP (in total, 102 constructs) was created. Allele-specific expression for 6 SNPs was detected by transfection of the HepG2 cell line with the constructed plasmid library and sequencing of target DNA and RNA sequences using the Illumina (MiSeq) platform.

Slight Variations in the Sequence Downstream of the Polyadenylation Signal Significantly Increase Transgene Expression in HEK293T and CHO Cells.

Compared to transcription initiation, much less is known about transcription termination. In particular, large-scale mutagenesis studies have, so far, primarily concentrated on promoter and enhancer, but not terminator sequences. Here, we used a massively parallel reporter assay (MPRA) to systematically analyze the influence of short (8 bp) sequence variants (mutations) located downstream of the polyadenylation signal (PAS) on the steady-state mRNA level of the upstream gene, employing an eGFP reporter and human HEK293T cells as a model system. In total, we evaluated 227,755 mutations located at different overlapping positions within +17..+56 bp downstream of the PAS for their ability to regulate the reporter gene expression. We found that the positions +17..+44 bp downstream of the PAS are more essential for gene upregulation than those located more distal to the PAS, and that the mutation sequences ensuring high levels of eGFP mRNA expression are extremely T-rich. Next, we validated the positive effect of a couple of mutations identified in the MPRA screening on the eGFP and luciferase protein expression. The most promising mutation increased the expression of the reporter proteins 13-fold and sevenfold on average in HEK293T and CHO cells, respectively. Overall, these findings might be useful for further improving the efficiency of production of therapeutic products, e.g., recombinant antibodies.

Statistical considerations for the analysis of massively parallel reporter assays data.

Noncoding DNA contains gene regulatory elements that alter gene expression, and the function of these elements can be modified by genetic variation. Massively parallel reporter assays (MPRA) enable high-throughput identification and characterization of functional genetic variants, but the statistical methods to identify allelic effects in MPRA data have not been fully developed. In this study, we demonstrate how the baseline allelic imbalance in MPRA libraries can produce biased results, and we propose a novel, nonparametric, adaptive testing method that is robust to this bias. We compare the performance of this method with other commonly used methods, and we demonstrate that our novel adaptive method controls Type I error in a wide range of scenarios while maintaining excellent power. We have implemented these tests along with routines for simulating MPRA data in the Analysis Toolset for MPRA (@MPRA), an R package for the design and analyses of MPRA experiments. It is publicly available at http://github.com/redaq/atMPRA.

Predicting functional variants in enhancer and promoter elements using RegulomeDB.

Here we present a computational model, Score of Unified Regulatory Features (SURF), that predicts functional variants in enhancer and promoter elements. SURF is trained on data from massively parallel reporter assays and predicts the effect of variants on reporter expression levels. It achieved the top performance in the Fifth Critical Assessment of Genome Interpretation "Regulation Saturation" challenge. We also show that features queried through RegulomeDB, which are direct annotations from functional genomics data, help improve prediction accuracy beyond transfer learning features from DNA sequence-based deep learning models. Some of the most important features include DNase footprints, especially when coupled with complementary ChIP-seq data. Furthermore, we found our model achieved good performance in predicting allele-specific transcription factor binding events. As an extension to the current scoring system in RegulomeDB, we expect our computational model to prioritize variants in regulatory regions, thus help the understanding of functional variants in noncoding regions that lead to disease.

Integration of multiple epigenomic marks improves prediction of variant impact in saturation mutagenesis reporter assay.

The integrative analysis of high-throughput reporter assays, machine learning, and profiles of epigenomic chromatin state in a broad array of cells and tissues has the potential to significantly improve our understanding of noncoding regulatory element function and its contribution to human disease. Here, we report results from the CAGI 5 regulation saturation challenge where participants were asked to predict the impact of nucleotide substitution at every base pair within five disease-associated human enhancers and nine disease-associated promoters. A library of mutations covering all bases was generated by saturation mutagenesis and altered activity was assessed in a massively parallel reporter assay (MPRA) in relevant cell lines. Reporter expression was measured relative to plasmid DNA to determine the impact of variants. The challenge was to predict the functional effects of variants on reporter expression. Comparative analysis of the full range of submitted prediction results identifies the most successful models of transcription factor binding sites, machine learning algorithms, and ways to choose among or incorporate diverse datatypes and cell-types for training computational models. These results have the potential to improve the design of future studies on more diverse sets of regulatory elements and aid the interpretation of disease-associated genetic variation.

Optimized PCR conditions minimizing the formation of chimeric DNA molecules from MPRA plasmid libraries.

BACKGROUND: Massively parallel reporter assays (MPRAs) enable high-throughput functional evaluation of various DNA regulatory elements and their mutant variants. The assays are based on construction of highly diverse plasmid libraries containing two variable fragments, a region of interest (a sequence under study; ROI) and a barcode (BC) used to uniquely tag each ROI, which are separated by a constant spacer sequence. The sequences of BC-ROI combinations present in the libraries may be either known a priori or not. In the latter case, it is necessary to identify these combinations before performing functional experiments. Typically, this is done by PCR amplification of the BC-ROI regions with flanking primers, followed by next-generation sequencing (NGS) of the products. However, chimeric DNA molecules formed on templates with identical spacer fragment during the amplification process may substantially hamper the identification of genuine BC-ROI combinations, and as a result lower the performance of the assays. RESULTS: To identify settings that minimize formation of chimeric products we tested a number of PCR amplification parameters, such as conventional and emulsion types of PCR, one- or two-round amplification strategies, amount of DNA template, number of PCR cycles, and the duration of the extension step. Using specific MPRA libraries as templates, we found that the two-round amplification of the BC-ROI regions with a very low initial template amount, an elongated extension step, and a specific number of PCR cycles result in as low as 0.30 and 0.32% of chimeric products for emulsion and conventional PCR approaches, respectively. CONCLUSIONS: We have identified PCR parameters that ensure synthesis of specific (non-chimeric) products from highly diverse MPRA plasmid libraries. In addition, we found that there is a negligible difference in performance of emulsion and conventional PCR approaches performed with the identified settings.

Experimental Evolution of Escherichia coli K-12 in the Presence of Proton Motive Force (PMF) Uncoupler Carbonyl Cyanide m-Chlorophenylhydrazone Selects for Mutations Affecting PMF-Driven Drug Efflux Pumps.

Experimental evolution of Escherichia coli K-12 with benzoate, a partial uncoupler of the proton motive force (PMF), selects for mutations that decrease antibiotic resistance. We conducted experimental evolution in the presence of carbonyl cyanide m-chlorophenylhydrazone (CCCP), a strong uncoupler. Cultures were serially diluted daily 1:100 in LBK medium containing 20 to 150 µM CCCP buffered at pH 6.5 or at pH 8.0. After 1,000 generations, the populations tolerated up to 150 µM CCCP. Sequenced isolates had mutations in mprA (emrR), which downregulates the EmrAB-TolC pump that exports CCCP. A mprA::kanR deletion conferred growth at 60 µM CCCP, though not at the higher levels resisted by evolved strains (150 µM). Some mprA mutant strains also had point mutations affecting emrA, but deletion of emrA abolished the CCCP resistance. Thus, CCCP-evolved isolates contained additional adaptations. One isolate lacked emrA or mprA mutations but had mutations in cecR (ybiH), whose product upregulates drug pumps YbhG and YbhFSR, and in gadE, which upregulates the multidrug pump MdtEF. A cecR::kanR deletion conferred partial resistance to CCCP. Other multidrug efflux genes that had mutations included ybhR and acrAB The acrB isolate was sensitive to the AcrAB substrates chloramphenicol and tetracycline. Other mutant genes in CCCP-evolved strains include rng (RNase G) and cyaA (adenylate cyclase). Overall, experimental evolution revealed a CCCP-dependent fitness advantage for mutations increasing CCCP efflux via EmrA and for mutations that may deactivate proton-driven pumps for drugs not present (cecR, gadE, acrAB, and ybhR). These results are consistent with our previous report of drug sensitivity associated with evolved benzoate tolerance.IMPORTANCE The genetic responses of bacteria to depletion of proton motive force (PMF), and their effects on drug resistance, are poorly understood. PMF drives export of many antibiotics, but the energy cost may decrease fitness when antibiotics are absent. Our evolution experiment reveals genetic mechanisms of adaptation to the PMF uncoupler CCCP, including selection for increased CCCP efflux but also against the expression of PMF-driven pumps for drugs not present. The results have implications for our understanding of the gut microbiome, which experiences high levels of organic acids that decrease PMF.

High-Throughput Analysis of Retinal Cis-Regulatory Networks by Massively Parallel Reporter Assays.

Inherited retinal degenerations are diverse and debilitating blinding diseases. Genetic tests and exome sequencing have identified mutations in many protein-coding genes associated with such diseases, but causal sequence variants remain to be found in many retinopathy cases. Since 99% of our genome does not code for protein but contains cis-regulatory elements (CREs) that regulate the expression of essential genes, CRE variants might hold the answer for some of these cases. However, identifying functional CREs within the noncoding genome and predicting the pathogenicity of CRE variants pose a significant challenge. Here, we review the development of massively parallel reporter assays in the mouse retina, its use in dissecting retinal cis-regulatory networks, and its potential application for developing therapies.

A distant trophoblast-specific enhancer controls HLA-G expression at the maternal-fetal interface.

HLA-G, a nonclassical HLA molecule uniquely expressed in the placenta, is a central component of fetus-induced immune tolerance during pregnancy. The tissue-specific expression of HLA-G, however, remains poorly understood. Here, systematic interrogation of the HLA-G locus using massively parallel reporter assay (MPRA) uncovered a previously unidentified cis-regulatory element 12 kb upstream of HLA-G with enhancer activity, Enhancer L Strikingly, clustered regularly-interspaced short palindromic repeats (CRISPR)/Cas9-mediated deletion of this enhancer resulted in ablation of HLA-G expression in JEG3 cells and in primary human trophoblasts isolated from placenta. RNA-seq analysis demonstrated that Enhancer L specifically controls HLA-G expression. Moreover, DNase-seq and chromatin conformation capture (3C) defined Enhancer L as a cell type-specific enhancer that loops into the HLA-G promoter. Interestingly, MPRA-based saturation mutagenesis of Enhancer L identified motifs for transcription factors of the CEBP and GATA families essential for placentation. These factors associate with Enhancer L and regulate HLA-G expression. Our findings identify long-range chromatin looping mediated by core trophoblast transcription factors as the mechanism controlling tissue-specific HLA-G expression at the maternal-fetal interface. More broadly, these results establish the combination of MPRA and CRISPR/Cas9 deletion as a powerful strategy to investigate human immune gene regulation.

Predicting enhancer activity and variant impact using gkm-SVM.

We participated in the Critical Assessment of Genome Interpretation eQTL challenge to further test computational models of regulatory variant impact and their association with human disease. Our prediction model is based on a discriminative gapped-kmer SVM (gkm-SVM) trained on genome-wide chromatin accessibility data in the cell type of interest. The comparisons with massively parallel reporter assays (MPRA) in lymphoblasts show that gkm-SVM is among the most accurate prediction models even though all other models used the MPRA data for model training, and gkm-SVM did not. In addition, we compare gkm-SVM with other MPRA datasets and show that gkm-SVM is a reliable predictor of expression and that deltaSVM is a reliable predictor of variant impact in K562 cells and mouse retina. We further show that DHS (DNase-I hypersensitive sites) and ATAC-seq (assay for transposase-accessible chromatin using sequencing) data are equally predictive substrates for training gkm-SVM, and that DHS regions flanked by H3K27Ac and H3K4me1 marks are more predictive than DHS regions alone.