Phase separation of ligand-activated enhancers licenses cooperative chromosomal enhancer assembly - PubMed (original) (raw)

. 2019 Mar;26(3):193-203.

doi: 10.1038/s41594-019-0190-5. Epub 2019 Mar 4.

Lu Yang 2, Dario Meluzzi 2, Soohwan Oh 2 3, Feng Yang 2, Meyer J Friedman 2, Susan Wang 2 4, Tom Suter 2, Ibraheem Alshareedah 5, Amir Gamliel 2, Qi Ma 2, Jie Zhang 2, Yiren Hu 2 3, Yuliang Tan 2, Kenneth A Ohgi 2, Ranveer Singh Jayani 2, Priya R Banerjee 5, Aneel K Aggarwal 6, Michael G Rosenfeld 7

Affiliations

Phase separation of ligand-activated enhancers licenses cooperative chromosomal enhancer assembly

Sreejith J Nair et al. Nat Struct Mol Biol. 2019 Mar.

Abstract

A crucial feature of differentiated cells is the rapid activation of enhancer-driven transcriptional programs in response to signals. The potential contributions of physicochemical properties of enhancer assembly in signaling events remain poorly understood. Here we report that in human breast cancer cells, the acute 17β-estradiol-dependent activation of functional enhancers requires assembly of an enhancer RNA-dependent ribonucleoprotein (eRNP) complex exhibiting properties of phase-separated condensates. Unexpectedly, while acute ligand-dependent assembly of eRNPs resulted in enhancer activation sensitive to chemical disruption of phase separation, chronically activated enhancers proved resistant to such disruption, with progressive maturation of eRNPs to a more gel-like state. Acute, but not chronic, stimulation resulted in ligand-induced, condensin-dependent changes in spatial chromatin conformation based on homotypic enhancer association, resulting in cooperative enhancer-activation events. Thus, distinct physicochemical properties of eRNP condensates on enhancers serve as determinants of rapid ligand-dependent alterations in chromosomal architecture and cooperative enhancer activation.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interests.

Figures

Fig. 1 |

Fig. 1 |. Acutely active E2-responsive MegaTrans enhancers concentrate a protein complex that can undergo phase transition.

a, Schematic representation of the ERα/MegaTrans complex recruited to E2-activated enhancers, which transcribe eRNAs and recruit the condensin I/II complexes. b, Western blot analyses showing that ERα, several MegaTrans components and condensin component SMC4 are precipitated by biotinylated isoxazole (b-isoxazole). FUS and GAPDH proteins are used as a positive and negative control for the assay, respectively. WCL, whole-cell lysate. c,d, FRAP data on phase-separated droplet formed in vitro by purified recombinant GATA3-maltose binding protein (MBP) (c) and ERα-MBP (d). Top, charts show individual data points represented by dots, lines represent fitting to an exponential model to estimate the half-time of recovery. Bottom, representative images of in vitro droplets before and after photobleaching. e, Fluorescence microscopy images of a representative nucleus from MCF7 cells transfected with ERα-mCherry, before (−E2) or after (+E2, 5 or 15min) E2 treatment. Scale bar, 2 μm. f, Mean intensity and photobleaching normalized fluorescence of ERα-mTurquoise foci in E2 treated MCF7 cells relative to pre-bleaching signal. Error bars represent s.e.m. of _n_≥10 cells per time point. g, Levels of eRNA from indicated enhancers, measured by reverse transcription PCR, from MCF7 cells depleted of endogenous GATA3 and expressing either wild type or IDR-deleted GATA3 (GATA3-IDR mut), after 1 h E2 stimulation. shCTL indicates non-targeting control shRNA. The IDR (aa 2–250) is shown in the schematics on top. Results are shown as individual data points (circles), mean±s.d. (lines). P values were calculated with an unpaired Student’s _t_-test. Data are representative of three independent experiments.

Fig. 2 |

Fig. 2 |. Effect of phase-separation inhibition on acute enhancer transcriptional activation.

a, Meta-analysis of genome-wide GRO-seq data for enhancer activity in cells treated with 2,5-HD or 1,6-HD and E2. Enhancers are classified as MegaTrans, weak ERα and non-ERα-bound enhancers. b-d, Meta-analysis of chromatin immunoprecipitation-(ChIP)-seq data representing the effect of 1,6-HD on chromatin recruitment of ERα (b), AP2y (c) and GATA3 (d) on MegaTrans enhancers. e, Left, representative fluorescence microscopy images from MCF7 cells expressing ERα-Turquoise, showing loss of E2-induced ERα foci on 1,6-HD treatment. Scale bars, 5μm. Right, quantification of foci number and intensity on 1,6-HD treatment. Foci number data are shown in bar graphs, as mean and s.e.m. of n = 3 cells. Foci intensity data are shown as box plots, in which boxes represent interquartile ranges (IQRs); the whisker represents points in lower and upper quartiles within 1.5 IQR from lower and upper edges of IQR. The data are for n = 539, 446 and 436 estrogen-receptor foci for pre-, post-2 min and post-4min 1,6-HD time points, respectively. P values calculated with a two-tailed _z_-test. f, Meta-analysis of GRO-seq data showing impact of 1,6-HD on TNFα (1 h) activation of p65-bound enhancers in MCF7 cells. g, Meta-analysis of GRO-seq data showing impact of 1,6-HD on KLA-stimulated (30 min) enhancers in RAW264.7 cells.

Fig. 3 |

Fig. 3 |. Rapid ligand-induced interactions between distant MegaTrans enhancers.

a, Schematic diagram of human chromosome 21 showing enhancers with the highest levels of ERα binding and transcriptional activation in MCF7 cells following 1 h E2 stimulation,. Active transcription units are listed below. *, >1 enhancer. b, Representative DNA FISH images showing the E2 induced proximity of indicated MegaTrans enhancer loci. Arrowhead points to the pair of loci in proximity. The TFF1 and DSCAM-AS1 loci are aneuploid in MCF7 cells, hence >2 FISH signals. Scale bars, 2 μm. c, Cumulative distribution of distances between indicated MegaTrans enhancers, with and without E2 stimulation. Data pooled from >200 nuclei from at least two independent experiments. P values were calculated using the Kolmogorov-Smirnov test. d, Time course quantifying the fold changes in fractions of TFF1/NRIP1 allele pairs with spatial separation <600 nm. Error bars are theoretical standard deviations and P values were calculated using a bootstrap method (see Methods). For each time point, more than 650 distances were pooled from at least two independent experiments. e, RNA FISH using NRIP1 and TFF1 intronic mRNA probes after 15min of E2 treatment, showing increased transcription from NRIP1 allele in proximity to TFF1 (indicated by arrows) compared to a allele that is spatially distant from TFF1 (indicated by arrowhead). f, Quantitation of RNA FISH signal intensity in relation to spatial distances between three different pairs of loci: NRIP1 in relation to TFF1 and to DSCAM-AS1; DOPEY2 in relation to TFF1. a.u., arbitrary unit. Boxes represent IQRs; whiskers represent points in lower and upper quartiles within 1.5 IQR from lower and upper edges of IQR. For each pair of loci, at least 76 data points were pooled from two independent experiments. P values were calculated using a Wilcoxon rank sum test with continuity correction. g, Cumulative distribution of distances between NRIP1 and TFF1 enhancers in MCF7 cells after 50 min E2 stimulation, treated with 2,5-HD or 1,6-HD. P values calculated using the Kolmogorov-Smirnov test. h, Cumulative frequency distribution of distances between E2-induced ERα-Turquoise foci and SC35-RFP (ICG marker).

Fig. 4 |

Fig. 4 |. Role of enhancer RNAs and condensins in E2 induced chromosomal dynamics and eRNP assembly.

a, ChIP data showing effect of TFF1e eRNA knockdown on recruitment of ERα, GATA3, RARα and AP2γ to the TFF1e enhancer following E2 treatment. CTL-ASO indicates non-targeting oligo used as control. Data shown as individual values (circles), mean and s.d. (lines). P values calculated with an unpaired Student’s _t_-test. Data are representative of three independent experiments. b, Fold changes in fractions of NRIP1/TFF1 allele pairs with separation below the cut-off distance, showing that TFF1e eRNA knockdown abolishes E2 induced proximity between TFF1 and NRIP1 enhancer loci. Error bars show theoretical sample standard deviations and P values were calculated using a bootstrap method (see Methods). CTL-ASO indicates a non-targeting oligo used as a control. c, d, FRAP kinetics showing the effect of TFF1 eRNA (0.20 mgml) on GATA3-MBP (c) or ERα-MBP or (d) fusion protein droplets in vitro. Data points are shown as circles, lines represent fitting to an exponential model to estimate the half-time or recovery. e, ChIP data showing effect of knocking down TFF1 eRNA or NRIPIe eRNA on recruitment of condensin II subunit NCAPH2 (top) or condensin I subunit NCAPG (bottom) to the TFF1e and NRIPIe loci. CTL-ASO indicates a non-targeting oligo used as a control. Data shown are individual values (circles), mean and s.d. (lines). P values were calculated with an unpaired Student’s f-test. Data are representative of three independent experiments. f, Fractions of NRIP1/TFF1 allele pairs with separation below the cut-off distance in cells with knockdown of the indicated proteins. Error bars indicate theoretical sample standard deviations and P values were calculated using a bootstrap method (see Methods). For each time point, more than 290 distances were pooled from at least two independent experiments. g, GRO-seq analyses showing effects of knockdown of NCAPG (condensin I) or NCAPD3 (condensin II) on E2-activated enhancer transcription. siCTL represent scrambled oligos used as control; results are grouped for MegaTrans, weak ERα enhancers and non-ERα-bound enhancers.

Fig. 5 |

Fig. 5 |. Chronic stimulation with E2 causes a fluid to hydrogel-like transition at enhancers and prevents ligand-induced enhancer proximity.

a, Box plots of GRO-seq analysis MCF7 cells not stimulated with E2 and not treated with 1,6-Hexanediol (1,6-HD) or E2 stimulation for 14h with or without treatment with 1,6-HD for 5 min. Central line shows median; boxes represent the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles. P values were calculated using two-tailed Student’s _t_-test. b, Fluorescent recovery of ERα-mTurquoise foci in MCF7 cells after short-term (30 min) or long-term (16 h) E2 treatment. Each point represents the mean intensity of the photobleaching normalized fluorescence relative to pre-bleaching signal. Error bars represent s.e.m. of n = at least 10 cells per time point. c, Box plots showing time constants of FRAP recovery of ERα-mTurquoise foci in MCF7 cells treated with E2 for long term (n = 28) and short term (n = 26). Boxes represent IQRs; the whiskers represent points in lower and upper quartiles within 1.5 IQR from lower and upper edges of IQR. P values were calculated with a two-tailed _z_-test. d, FRAP analysis of in vitro droplets formed by a ternary mixture of GATA3-MBP (7μM), ERα-MBP (7μM) and TFF1 eRNA (0.20 mgml−), and incubated for 5, 90 or 180 min. Data show a less rapid fluorescence recovery with increasing time of incubation. Data points are represented by dots, lines represent fitting to an exponential model e, Cumulative distribution of distances between indicated MegaTrans enhancer pairs after short-term (50 min) and long-term (16 h) E2 treatment, demonstrating that the E2-induced spatial proximity is lost on prolonged treatment. Data pooled from >150 nuclei examined in two independent experiments. P values were calculated using the Kolmogorov-Smirnov test.

Fig. 6 |

Fig. 6 |. Model summary.

MegaTrans enhancers are minimally active under unstimulated conditions. E2 stimulation results in ERα dependent recruitment of MegaTrans complex, condensin complex and eRNA transcription, forming ‘megaloops’ between these enhancers. This results in an eRNA-dependent RNP (eRNP) assembled by phase separation. Chronic stimulation alters the physicochemical properties of this complex to a ‘gelsol state’ thus making them less sensitive to 1,6-HD. Maximal enhancer activation occurs with colocalization of the enhancer in the ICG, apparently resulting in increased concentration of transcriptional machinery and increased transcriptional robustness of component enhancers. Association with other nuclear structures such as the nucleolus and nuclear lamina represent a speculative model based on A/B compartments and rDNA locus in human chromosome 21.

Comment in

References

    1. Bulger M & Groudine M Functional and mechanistic diversity of distal transcription enhancers. Cell 144, 327–339 (2011). - PMC - PubMed
    1. Long HK, Prescott SL & Wysocka J Ever-changing landscapes: transcriptional enhancers in development and evolution. Cell 167, 1170–1187 (2016). - PMC - PubMed
    1. Loven J et al. Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell 153, 320–334 (2013). - PMC - PubMed
    1. Whyte WA et al. Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell 153, 307–319 (2013). - PMC - PubMed
    1. Parker SC et al. Chromatin stretch enhancer states drive cell-specific gene regulation and harbor human disease risk variants. Proc. Natl Acad. Sci. USA 110, 17921–17926 (2013). - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources