Figure 6.

Modeling of chromatin states reveals the dynamics of LADs and GADs during adipogenic differentiation. (A) ChromHMM emission parameters: heat map of the relative abundance of chromatin states (numbered 1–15) in each indicated chromatin mark. The four lamin A/C-containing states are labeled green (states 2, 3, 4, 15). (B) Heat map of the relative abundance of the 15 states on predefined genomic regions on D0 and D3 of differentiation. Distribution of the four lamin A/C-containing states is shown in green areas in D0. On D3, red and yellow areas depict significant reductions (red) or increases (yellow) in enrichment levels of the lamin A/C-containing states. D−2, D0, D3, and D9 time point data and statistics are shown in Supplemental Figure 6A,C. (C) A two-step model of formation of lamin A/C LADs during adipogenic differentiation. In proliferating adipocyte progenitors, LAD coverage is relatively limited and does not necessarily involve GADs. After cell-cycle arrest, a necessary step for adipogenic differentiation, LAD coverage is extended independently of GADs. In undifferentiated cells, lamin A/C LADs contain both repressed chromatin, presumably harboring H3K9me2/3 (Guelen et al. 2008; Kind et al. 2013; Harr et al. 2015) and active chromatin domains. Adipogenic differentiation elicits an exchange of lamin A/C LADs; this involves the formation of LADs predominantly on H2BGlcNAc domains, consistent with an epigenetic prepatterning of de novo adipogenic lamin A/C LADs by GADs. The recently shown involvement of H3K27me3-enriched regions in LAD borders on the maintenance of LADs in mouse cells (Guelen et al. 2008; Kind et al. 2013; Harr et al. 2015) raises the possibility that de novo lamin A/C LADs formed during adipogenesis also entail a contribution from trimethylated H3K27 in LAD borders. The overall repressed state of these de novo LADs, together with their strong overlap with lamin B1 LADs, suggests that they become enriched in heterochromatin marked by di- or trimethylated H3K9.