Data obtained from analysing chromosomal organization and interactions in individual cells unify previous results obtained by single-cell imaging and studies of population-averaged genomic interactions. See Article p.59
Each human cell carries DNA molecules that, when combined, are more than two metres long. A fascinating problem in cell biology is, therefore, how these long molecules are organized inside the cell's nucleus. This question is not just of interest from a structural perspective. Understanding chromosome folding will also provide deeper insight into genomic processes, such as the regulation of gene expression and the maintenance of genome stability. On page 59 of this issue, Nagano et al.1 describe a genomic approach to probing the three-dimensional arrangement of chromosomes within single cellsFootnote 1. The results reveal large cell-to-cell variation in chromosome structure and nuclear organization, but also show that common principles of organization derived previously from cell-population studies apply to individual cells.
Chromosomes and individual genomic regions (loci) have been studied in single cells using microscopy2. Such work has shown that although chromosomes are not randomly organized, their structure varies between cells in a population. However, approaches at the molecular level, based on the chromosome conformation capture (3C) technique, have produced maps of averaged genome-wide interactions in chromatin (DNA–protein complexes)3, allowing chromosome organization across large cell populations to be determined. These studies have led to the discovery that there are general principles of chromosome folding4,5, but they give no information about cell-to-cell variability, nor how these folding principles are implemented at the single-cell level. Nagano et al. present a variant of the 3C technique (called single-cell Hi-C) that allows genome-wide measurement of chromatin interactions in individual cells.
In 3C, cells are chemically fixed to covalently link any pair of genomic loci that are in close spatial proximity. The chromatin is then digested with a restriction enzyme, and DNA ends are joined to form unique DNA-ligation products, each of which represents a contact between two loci in a cell in the population. The resulting ligation-product library can then be interrogated using deep-sequencing methods, to determine the genomic location of interacting loci (the Hi-C approach)6. Because 3C methods are typically performed on millions of cells, the measured chromatin-interaction frequencies reflect the probability with which loci are interacting across the population. However, it is difficult to use the data generated to discover which interactions occur together in single cells, how many cells display a particular pair-wise interaction, or to what extent nuclear organization varies in different cells.
Nagano and colleagues' single-cell Hi-C method adds a clever twist to the 3C procedure. It involves most of the steps of the classical 3C technique, but is performed inside the nuclei of permeabilized cells. Subsequent isolation of single nuclei allows all DNA-ligation products to be collected from each of the cells. Deep sequencing of these products yields a list of pair-wise chromatin interactions from that cell. The authors further applied statistical analyses to extract significant patterns of chromatin interaction that reflect various features of the genome's spatial organization.
Analysing individual immune cells called T helper cells, Nagano et al. note a striking level of cell-to-cell variability in interactions between and within chromosomes. Whereas in one cell a chromosome (more strictly, two homologous chromosomes) might interact with almost all others, in another cell the same chromosome might interact with only a few others. In part, this is consistent with earlier microscopy observations2 showing that each chromosome occupies a distinct 'chromosomal territory' and interacts with only a few neighbours, which differ from cell to cell.
How can the ability of a chromosome to interact with a dozen other chromosomes be reconciled with the formation of chromosomal territories? It could be that territory shape is highly irregular, with sub-chromosomal domains protruding from the main chromosomal mass. High-resolution microscopy should allow this hypothesis to be tested. It also remains to be seen whether the propensity of chromosomes to engage in many or only a few trans-chromosomal interactions reflects a state of individual chromosomes, or that of an individual cell — for example, its progression through the cell cycle or its response to signals.
Nagano et al. also report cell-to-cell variation in the internal organization of chromosomes. However, when disparate interaction patterns from individual cells are pooled, a common pattern emerges — the interactions 'clump' together. Such clumps correspond to interactions within previously identified topologically associating domains (TADs)7,8,9, which were first discovered using population-averaged chromatin-interaction maps. TADs have been proposed to form the building blocks for modular assembly of larger-scale structures10. This modular architecture is also evident in trans-chromosomal interactions, which are mostly formed not by isolated, protruding genomic regions, but by entire domains. Nagano and co-workers' Hi-C data are too sparse to determine whether every TAD is present in every cell. But their statistical analysis suggests that cell-to-cell variability in chromosome organization is partly due to differential assembly of relatively reproducible, modular sub-chromosomal domains.
Although interactions between the structural domains differ among cells, they are not random, instead reflecting functional states of individual domains. Consistent with population-averaged data6,11,12, in individual cells, domains enriched for active genomic regions tend to interact with other active domains; similarly, inactive domains tend to interact with other inactive domains (Fig. 1).
The information on interactions within a single chromosome can be used to build a three-dimensional model of that chromosome and so reproduce the observed interactions. Nagano et al. build three-dimensional models of the single-copy X chromosome (to eliminate ambiguity caused by homologous chromosomes) for several individual cells. The models show that domains marked by active chromatin are frequently located near the periphery of a chromosome, where they may interact with active domains on other chromosomes, in agreement with population-averaged data11.
Population-based data have also suggested that the probability of interactions between a pair of loci on the same chromosome decreases approximately as the inverse of the genomic distance between the loci. Despite cell-to-cell variation in specific interactions, scaling of the interaction probability with genomic distance is surprisingly consistent between individual cells and agrees with the population-based scaling.
Questions remain. For example, how variable is chromatin organization within the modular chromosome domains? Are interactions between genes and their regulatory elements also stochastic across the cell population? What is the contribution of real-time chromatin dynamics to cell-to-cell variation, and is this different for local structures and higher-order conformations?
Answering some of these questions will require further technological developments, for instance to allow the capture of more than the currently possible 2.5% of all interactions in a cell. Answering others may require a combination of assays within the same cell — such as single-cell Hi-C with analysis of all RNA transcripts. This should help to uncover how cell-specific chromosome conformation relates to stochastic gene expression. Single-cell Hi-C promises to become an important approach in determining chromosomal organization within cells and its relevance to gene expression.
*This article and the paper under discussion1 were published online on 25 September 2013.
Nagano, T. et al. Nature 502, 59–64 (2013).
Cremer, T. & Cremer, M. Cold Spring Harb. Perspect. Biol. 2, a003889 (2010).
Dekker, J., Rippe, K., Dekker, M. & Kleckner, N. Science 295, 1306–1311 (2002).
Bickmore, W. A. & van Steensel, B. Cell 152, 1270–1284 (2013).
Dekker, J., Marti-Renom, M. A. & Mirny, L. A. Nature Rev. Genet. 14, 390–403 (2013).
Lieberman-Aiden, E. et al. Science 326, 289–293 (2009).
Dixon, J. R. et al. Nature 485, 376–380 (2012).
Nora, E. P. et al. Nature 485, 381–385 (2012).
Sexton, T. et al. Cell 148, 458–472 (2012).
Gibcus, J. H. & Dekker, J. Mol. Cell 49, 773–782 (2013).
Kalhor, R., Tjong, H., Jayathilaka, N., Alber, F. & Chen, L. Nature Biotechnol. 30, 90–98 (2011).
Yaffe, E. & Tanay, A. Nature Genet. 43, 1059–1065 (2011).
About this article
The Dm-Myb Oncoprotein Contributes to Insulator Function and Stabilizes Repressive H3K27me3 PcG Domains
Cell Reports (2020)
3DeFDR: statistical methods for identifying cell type-specific looping interactions in 5C and Hi-C data
Genome Biology (2020)
Fluorescence fluctuation spectroscopy: an invaluable microscopy tool for uncovering the biophysical rules for navigating the nuclear landscape
Biochemical Society Transactions (2019)
The Dm-Myb Oncoprotein Coordinates Higher-Order Chromatin Structure and Stabilizes Facultative Heterochromatin
SSRN Electronic Journal (2019)
Microbiology and Molecular Biology Reviews (2015)