High-Precision Determination of Linear and Three-Dimensional Genomes of Single Cells

Abstract

In human and many other metazoans, both linear genome sequence and three-dimensional nuclear structure can vary between different cells within the same organism. Variations in linear DNA reflects the mutational history of the cell, and those in three-dimensional genome folding help regulate gene expression. Single-cell genomic assays therefore provide valuable means for characterizing the composition and regulation of genomes on a cell-to-cell basis. A sensitive and reproducible method for amplifying the genome of each single cell is the first step for decoding its linear DNA. In Chapter 2 we describe our single-cell whole-genome amplification assay that utilizes droplet microfluidics to produce long amplicons with accurate allele counting capability. This method has reproducibly achieved high genomic coverages of 91.0% ± 3% (average ± standard deviation) for all the cells verified to be non-replicated. To prevent DNA self-looping, we further included 16 transposon sequences, and labeled the two complementary strands of double-stranded DNA to correct for artifacts in the detection of single-nucleotide variations (SNVs). Described in Chapter 3, this assay, termed Multiplexed End-Tagging Amplification (META) of Complementary Strands, has revealed the mutational histories of blood and neuron cells with the highest accuracy known so far. Contrary to the conventional understanding that cell types are characterized by their transcriptomes, we found that SNV patterns are cell-type dependent and can be used to define cell types by principle component analysis. By integrating META with chromosome conformation capture, we re-constructed the genome-wide nuclear structures of single diploid human cells to an unprecedented resolution of 20kb. In Chapter 3 our method of Diploid Chromosome Conformation Capture (Dip-C) is illustrated; it has uncovered the cell-type dependent 3D genome architecture for human peripheral blood mononuclear cells. For the first time, the Waddington epigenetic landscape of human cells can be determined based on 3D genome folding, with valleys representing cell types and the width of each valley specifying cell-to-cell differences. Survey into more cell types at higher spatial resolution will help elucidate the relationships between genome structures and their functions in health and disease.