Integrated microfluidic tools for the characterization of protein/DNA interactions in vitro and in vivo

The specific interaction between DNA and proteins constitutes one of the crucial elements in the regulation of gene expression. This thesis focuses on the development and optimization of two microfluidic-based technologies called SMiLE-seq and FloChIP that enable the sensitive and high-throughput analysis of two important aspects of protein/DNA interactions: respectively, 1) the transcription factor DNA-binding specificities and 2) the genome-wide distribution of chromatin-associated proteins.

In Chapter 1, we describe the core motivation, operating principles, optimization and results obtained with SMiLE-seq. As opposed to existing solutions, SMiLE-seq offers the possibility to screen a large library of randomized DNA for sequence-specific protein ligands in a miniaturized context. SMiLE-seq integrates in a single microfluidic chip the advantages of both HT-SELEX and MITOMI. We first demonstrate as proof-of-principle that SMiLE-seq successfully recapitulates with robustness and reproducibility the binding specificities of known factors belonging to different species and TF families. Subsequently, we target the numerous although largely uncharacterized family of TFs called KRAB-ZFPs. We initially set out to redesign the microfluidic and protocol architecture in order to attain maximal throughput and sensitivity. Next, we add the ability of probing the sensitivity of these factors to methylation by synthesizing randomized methylated DNA libraries. Finally, we proceed to test 101 KRAB-ZFPs with both methylated and non-methylated DNA. We obtain high confidence motifs for 43 factors, of which 22 we not sensitive to methylation, 10 yielded motifs only with methylated DNA and 10 only with non-methylated DNA. By integrating our SMiLE-seq data with published ChIP-exo data and in silico predictions, we develop a framework for systematically identifying which zinc fingers directly contribute to DNA binding for a given KRAB-ZFPs.

In Chapter 2, we describe the development and results obtained with FloChIP, a microfluidic implementation of the widespread ChIP-seq and sequential ChIP-seq protocols. FloChIP encompasses three main technological advances: 1) the multi-layered serface chemistry that allows any off-the-shelf antibody to be immobilized on-chip and and 2) a micropillar architecture that provides high surface-to-volume ratio and efficient removal of non-specific DNA. 3) The direct on-chip tagmentation of immunoprecipitated DNA. We validate our approach by first performing H3k27ac ChIP-seq on different number of sample inputs, i.e. from 500 to 10^6 cells. Next, we proceed to prove the flexibility of our approach by performing ChIP-seq on different histone marks – namely H3k27ac, H3k27me3, H3k4me3, H3k4me1 and H3k9me3 – and comparing them to existing ENCODE data. Both region-specific profiles and enrichment scores show high similarity with publicly available data. Subsequently, we set out to demonstrate the feasibility of on-chip sequential-ChIP-seq by performing consecutive H3k4me3 and H3k27me3 IP on chromatin derived from mouse embryonic stem cells. Finally, we illustrate the high-throughput feature of our technology by performing MEF2A-ChIP-seq on chromatin derived from 32 different lymphoblastoid cell lines.