Comprehensive screens and a universal zinc finger model enable transcription factor reprogramming 2 factor reprogrammingComprehensive screens and a universal zinc finger model enable transcription factor reprogramming

Programmable regulation of gene expression promises to be a powerful therapeutic venue for diseases caused by the misexpression of genes, haploinsufficiencies, or gain of function mutations. While dCas9-effector domain fusions have been used to alter expression levels, their in vivo application is limited by size and immunogenicity. Conversely, the Cys2His2 Zinc Finger domain offers ideal characteristics for in vivo use. However, their intricate engagement with the DNA has made the design of Zinc Finger arrays challenging. We here describe the screening of 49 billion protein-DNA interactions and development of the first deep learning model that solves Zinc Finger design for any genomic target. We demonstrate the versatility of designed Zinc Fingers as nucleases and as activators or repressors by seamlessly reprogramming human transcription factors. Overall design: Samples are synthetic zinc fingers fused to either an effector domain. Two replicates are included per sample and the input DNA was sequenced for each sample. As a positive control/reference, the YY1 zinc finger protein was used.