Table 2:

A detailed overview of select MPRA and CRISPR-based studies discussed in this review.

Method- Family	Method	Parallel Methods	Variant Selection Criteria	Targets	Context	Cell Lines	Cell Type	Application	Limitations	Secondary Validation	Source
MPRAs	Mutagenesis	NA	Random nucleotide substitutions in enhancers at a rate of 10% per position.	27000	Episomal	HEK293T	human kidney	Direct comparison of hundreds of thousands of putative regulatory sequences in a single cell culture.	Depends on (1) careful design of the sequence library, (2) minimization of artifacts during amplification and cloning, (3) high transfection efficiency, (4) and necessary power to detect transcriptional shifts.	NA	Melnikov et al 2012
	Mutagenesis	NA	Tested 2104 WT sequences & 3314 engineered variants with motif disruptions.	>5,000	Episomal	HepG2, K562	Carcinoma LL	(1) Manipulations of a large number of enhancers and disruptions for individual cis-regulatory motifs. (2) Well-suited to systematic testing of pairs or sets of elements, and de novo enhancer design.	(1) Unable to determine relative contribution of chromatin vs. primary sequence information (2) focused on distal enhancers at least 2 kb from any annotated TSS for the SV40 promoter region only	Luciferase validation	Kheradpour et al 2013,
	Mutagenesis	NA	20 promoter/enhancers (600bp loci)	30,000 SNPs/	Episomal	HepG2	Liver	(1) Scaled saturation mutagenesis to measure regulatory consequences of tens-of-thousands of regulatory elements (2) Longer sequences than are typical for MPRAs, up to 600 bp to provide more context.	(1) Limited with respect to context, both cis and trans, (2) reproducibility of measurements for elements with lower basal activity	RNAi	Kircher et al 2019,
	GWAS-based MPRA	NA	Selection of variants in high LD with 75 GWAS hits from red blood cell (RBC) traits	2,756 SNPs	Episomal	K562	Erythriod	Identified 32 functional variants representing 23 of the original 75 GWAS hits	(1) Not configured to detect functional variants in haplotypes that may be jointly causal and fall within more than one regulatory element. (2) Primarily as a screen to reduce set of leads.	Cas9 genome editing	Ulirsch et al 2016,
	GWAS-based MPRA	NA	1,049 SZ and 30 AD variants in high LD with lead SNPs from 64 and 9 GWSIG loci respectively	1228 SNPs	Episomal	K562, SK-SY5Y	LCLs	Identified 148 variants showing allelic differences in K562 and 53 in SK-SY5Y cells	(1) High potential for false negatives due to lack of native context. (2) False positives possible if regulatory effect on the reporter gene comes from other parts of the construct or if variant resides in closed chromatin.	NA	Myint, et al., 2020,
	eQTL-based MPRA	NA	Candidate eQTLs from RNAseq dataset of lymphoblastoid LCLs	3,642 cis-eQTLs	Episomal	HepG2, NA12878NA19239	Liver, LCLs	(1) Identified 842 eQTLs with a significant transcriptional shift between alleles. (2) Provides a discovery tool for linking a genetic locus to a phenotype.	(1) Cannot test for causality. (2) Endogenously silenced sequences explain a proportion of reported active sequences. (3) Positive predictive value of 34%–68%.	CRISPR-mediaterd allelic rep-placement	Tewhey et al 2016,
	LentiMPRA	RNA-seq, ATAC-seq, H3K27ac ChIP-seq	Identified by RNA-seq/ATAC/H3K27Ac/ChIP-seq bease on genes involved in neural differentiation	~ 2300	Chromosomally Integrated	HepG2, hESC	liver, hNPCs	Functional characterization of >1,500 temporal enhancers (1) lentiMPRA used in an episomal or integrated context (2) can be used in a wide variety of cell types (3) numerous barcodes per variant; and (4) extensive predictive modeling.	(1) Even as an integrated reporter assays, each tested element is removed from its native sequence location and epigenetic context	CRISPRi	Inoue et al 2017, 2019
	SuRE MPRA	Dnase-seq, ATAC-seq, H3K27ac	Randomly generated two SuRE libraries of ~300 million random fragment-barcode pairs	5.9 million SNPs	Episomal	K563, HepG2	Eryhriod, Liver	(1) Increased traditional MPRA scale by >100 folde. (2) Provides a resource to help identify causal SNPs among candidates generated by GWAS and eQTL studies	(1) Random SNPs assayed outside of endogenous context (3) Power to detect transcriptional shift may be limited by number of barcodes per fragment	CRIPSR/ Cas9 SNP editing	van Arensbergen et al. 2019
	two-Stage MPRA screen	NA	Random genome-wide DNA fragments	32,776 substitutions	Episomal	hESC	hNSCs	(1) MPRA in human neural stem cells. (2) Identified 532 HARs and HGEs with human-specific changes in enhancer activity in human neural stem cells.	(1) Effects were modest and lacked genomic context.	CRISPRi enhancer validation	Uebbing et al. 2020
CRISPR	Multiplexed CRISPRi, eQTL-inspired framework	scRNA-seq	Top 5,000 intergenic open chromatin regions in K562s	5,920	Endogenous	K562	LCLs	Identified 664 cis enhancer-gene pairs enriched for specific transcription factors, non- housekeeping status, and genomic and 3D conformational proximity to their target genes	(1) Not all enhancers susceptible to perturbation; (2) variable degree of gRNAs targeting ability; (3) enhancers may be required for initial establishment rather than maintenance. (4) not a comprehensive survey of noncoding landscape	CRISPRi singleton experiments	Gasperini et al. 2019
	Genome-Wide CRISPRa screen	NA	Library targeting all computationally predicted TFs a& other DNA-binding factors (TRANSFAC)	2,428	Endogenous	CamES	Neuron	(1) Systematically identify transcription factors that efficiently promote neuronal fate from ESCs. (2) Generated a quantitative GI map for the neuronal fate decision by pairwise activation of core neuronal-inducing factors.	(1) Scalability of the CRIPRa screens	Flow cytometry and cDNA expression.	Liu et al 2019
	CRISPRi-FlowFISH	ChIP-seq, Hi-C	Selected all DNase I hypersensitive (DHS) elements in K562 cells within 450 kb of 30 genes in five genomic regions	4,662	Hi-C and ChIP-seq	K562	LCLs	(1) Tests noncoding regulatory elements by mapping and modeling promoter–promoter regulation, functions of CTCF sites, and combinatorial effects. (2) Potential application to any gene. (3) Method uses endogenous genes allowing candidate target genes to be identified.	(1) Does not profile effects of intronic enhancers; (2) performance may decrease by weakly expressed genes	ABC prediction model	Fulco et al. 2019
	Pooled CRISPR screens	Parallel screens of enhancers, genes, & genomic background	Identified from H3K27sc datasets from human cortex, developing cortex, limb, embryonic stem cells, and adult tissues	10674 genes & 2,227 enhancers	Endogenous	H9 hESCs	hNSCs	(1) Probed gene disruptions affecting proliferation in model of human corticogenisis and their associations with neurodevelopmental disease	(1) sgRNA-Cas9 screening using Cas9 original method resulting in insertions, deletions, and substitutions. influencing the DNA directly versus inhibiting or activating expression or enhancer activity.	Confirm enhancer-gene interaction using Hi-C data	Geller et al. 2019
	Pooled genome- wide & CRISPRi screens	CROP-seq, longitudinal imagining	CRISPRi v2 H1 library with top 5 sgRNAs per gene (Horlbeck et al., 2016)	18,905 genes	Endogenous	hiPSCs	Neuron	(1) Identified distinct neuronal roles for ubiquitous genes; (2) an inducible and reversible method enabling the time-resolved dissection of human gene function; (3) perturbs gene function via partial knockdown; (4) longitudinal imaging provides timeline of toxicity and reveals gene-specific temporal patterns.	(1) Scalability limited by us of lentivirus, synthetic sgRNAs in arrayed CRISPRi screens would increase scalability; (2) False-positive phenotypes possible due to interference with the differentiation process.	Secondary CRISPR screens, CROP-seq	Tian et al 2019
	Pooled CRISPRa/i screen	CROP-seq	(1) Genome-wide survival-based screen. (2) Secondary CRISPR screens. (3) Crop-seq	Genome-wide and targeted screens > 1000 genes	Endogenous	hiPSCs	Neuron	(1) The first genome-wide CRISPRa/i screens in human neurons (2) Uncovered neuron-response pathways to chronic oxidative stress implicated in neurodegenerative disease (3) Established CRISPRbrain resource to compare gene function across human cell types	(1) CRISPR a/i screening protocol may confound results related to oxidative stress as gRNA integration could affect cellular stress	Lipomics in KO neurons	Tian et al. 2020

CC = cervical cancer cells, hESCs = Human Embryonic Stem Cells, HiPSCs = Human-induced Pluripotent Stem Cells, LCL = Lymphoblastoid cell lines, ML = myelogenous leukemia cells, hNSCs – Neural Stem cells, NPCs = Neuronal Progenitor Cells, CREST-seq= “cis-regulatory element scan by tiling-deletion and sequencing”