3D genomics across the tree of life reveals condensin II as a determinant of architecture type

Abstract

We investigated genome folding across the eukaryotic tree of life. We find two types of three-dimensional (3D) genome architectures at the chromosome scale. Each type appears and disappears repeatedly during eukaryotic evolution. The type of genome architecture that an organism exhibits correlates with the absence of condensin II subunits. Moreover, condensin II depletion converts the architecture of the human genome to a state resembling that seen in organisms such as fungi or mosquitoes. In this state, centromeres cluster together at nucleoli, and heterochromatin domains merge. We propose a physical model in which lengthwise compaction of chromosomes by condensin II during mitosis determines chromosome-scale genome architecture, with effects that are retained during the subsequent interphase. This mechanism likely has been conserved since the last common ancestor of all eukaryotes.

Figure 1.. A comprehensive overview of genome organization across evolution

Aggregate chromosome analysis (ACA) on Hi-C maps of 24 species. ACA involves the scaling of chromosome arms to a uniform length and aggregating the signal of all intra- and inter-chromosomal contacts. The depicted species represent three kingdoms: animals (yellow), fungi (blue) and plants (green), whose evolutionary relationship is presented in the phylogenetic tree. Each corner shows an example ACA map and a schematic drawing of one of the four chromosome-scale features. Presence of the condensin II subunits in each species is indicated by solid black circles. Left to right: SMC2/SMC4/CAP-H2/CAP-G2/CAP-D3.

Figure 2.. Condensin II prevents centromeric clustering and keeps apart heterochromatin domains

(a) Hi-C matrices of the depicted genotypes in Hap1 cells. (b) Immunofluorescence of centromeres (CREST) and DNA (DAPI), as quantified in (c). (d) Difference in DamID score relative to distance to centromere. Zoom-in includes 95% confidence interval of the mean in grey. (e) Immunofluorescence of centromeres (CREST), nucleoli (Nucleolin) and DNA (DAPI) (f) Quantification of the fraction of centromere intensity within 0.4 μm of nucleoli as shown in (e). (g) Immunofluorescence of centromeres (CenpA), heterochromatin (H3K9me3) and DNA (DAPI), as quantified in (h).

Figure 3.. Massive 3D genome changes hardly affect gene expression

(a) Gene expression of wild type relative to ΔCAP-H2. Unaffected genes depicted in grey, upregulated genes in blue and downregulated in red. (b) Number of genes in each category. (c) Percentage of active genes overlapping with LADs. (d) Intersection of differences in gene expression with differences in lamina association, depicting active genes within LADs. (e) Schematic model of centromeres (red) moving to the inner nucleus, and silenced genes that now localize to the lamina.

Figure 4.. Centromeric clustering is counteracted by lengthwise compaction and requires mitosis-to-interphase transition

(a) Quantification of centromeric foci before or after mitotic progression with/without condensin II. FACS plots depict cell cycle stages. Outliers (>60) were truncated and depicted as squares. (b) Example images of G1 cells as quantified in (a). (c-g) Simulation modeling using ten polymer chains as chromosomes. (c) Number of centromere clusters upon varying lengthwise compaction. ‘WT’ and ‘ΔC’ correspond to higher and lower lengthwise compaction, best recapitulating the experimental data observed in wild type and ΔCAP-H2 cells. Top: representative models for both states. (d) Representative simulation snapshots depicting ten chromosomes in different colors. (e) Quantification of the ratio of cis-contacts. (f) Simulated Hi-C matrices depicting contacts between the respective chromosomes. (g) Quantification of the proportion of trans-centromeric contacts.