Massively parallel reporter assay reveals promoter-, position-, and strand-specific effects in transcription start sites

Massively parallel reporter assays (MPRA) are a high-throughput method of assessing the activity of candidate cis-regulatory sequences, and can be used to detect allelic differences at disease-associated variants. Previous MPRA studies have screened thousands of functional SNPs associated with various complex traits and conditions. Most MPRA libraries utilize a single plasmid configuration, a single minimal promoter to drive expression, and a single-strand orientation, which may fail to capture the context-dependent activity of disease-associated cis-regulatory elements. We interrogate the potential regulatory differences introduced by variable MPRA plasmid promoters and positions. We used an MPRA library to quantify the activity of 1,305 pancreatic islet-derived transcription start sites generated from CAP analysis of gene expression profiling. We cloned fragments upstream or downstream of a reporter gene along with either the human insulin (INS) promoter or a synthetic housekeeping promoter (SCP1). We used elastic net regression to predict position-specific fragment activity based on enrichment of transcription factor binding site motifs, and generalized linear models to predict position-specific fragment activity from tissue-specific chromatin state regulatory annotations. Our results support the use of MPRA strategies that account for context-dependent factors when assaying candidate regulatory elements in pursuit of understanding complex genetic diseases.