Enhancer switching in cell lineage priming is linked to eRNA, Brg1's AT-hook, and SWI/SNF recruitment

Abstract

RNA transcribed from enhancers, i.e., eRNA, has been suggested to directly activate transcription by recruiting transcription factors and co-activators. Although there have been specific examples of eRNA functioning in this way, it is not clear how general this may be. We find that the AT-hook of SWI/SNF preferentially binds RNA and, as part of the esBAF complex, associates with eRNA transcribed from intronic and intergenic regions. Our data suggest that SWI/SNF is globally recruited in cis by eRNA to cell-type-specific enhancers, representative of two distinct stages that mimic early mammalian development, and not at enhancers that are shared between the two stages. In this manner, SWI/SNF facilitates recruitment and/or activation of MLL3/4, p300/CBP, and Mediator to stage-specific enhancers and super-enhancers that regulate the transcription of metabolic and cell lineage priming-related genes. These findings highlight a connection between ATP-dependent chromatin remodeling and eRNA in cell identity and typical- and super-enhancer activation.

Keywords: Brg1/Smarca4; EpiSCs; MLL3/4; RNAPII; SWI/SNF; eRNA; enhancer; mediator (Med1); naive; p300/CBP; primed.

Figure 1.. The AT-hook of Brg1 is required for stage-specific SWI/SNF recruitment.

(A) Mouse embryonic stem cells (E14Tg2a) are cultured with MEK and Gsk3b inhibitor for naïve stage or with FGF2 and Activin-A for primed stage cells. Binding of wild type Brg1 is detected by the CUT&RUN method and the genomic distribution of Brg1 peaks in naïve (left) and primed (right) stages are shown. (B) The number of overlapping peaks between Brg1 CUT&RUN and ATAC-seq is shown for those peaks that are naïve only (left), primed only (middle), or present in both (right). (C-E) Binding of wild type Brg1 is detected by the CUT&RUN method and peaks separated into naïve-specific (C, left panel), primed-specific (D, left panel), and shared (E, left panel). Brg1 peaks (blue) are shown with corresponding accessible peaks from ATAC-seq (grey) at intronic-intergenic sites for wild type (WT) and AT-hook deleted mutant (ΔAT1, ΔAT2) Brg1 in naïve (C), primed (D), and shared (E) sites. Heatmaps are sorted based on WT Brg1 in each stage. Total number analyzed peaks in naïve, primed, and shared sites are 40,909, 40,124, and 26,172 respectively. Rep1 and Rep2 are two biological replicates.

Figure 2.. The AT-hook of Brg1 preferentially binds RNA and transcription promotes Brg1 binding to stage-specific enhancers.

(A) Purified maltose binding protein (MBP)-wild type AT-hook and -mutant AT-hook fusion proteins are analyzed on 12% SDS-PAGE and stained with Coomassie blue. (B) The binding affinity of wild type (WT-AT) and mutant AT-hook (Mut-AT) fused to MBP are determined by electrophoretic mobility shift assays (EMSA) on a 6% native PAGE for Cy5 labeled 7SK-L2-RNA (red) and Cy3 labeled mouse rDNA enhancer (green). Reactions contained 0.03, 0.1, 0.2, 0.4, 0.8, 1.6, 5.0, and 20.0 μM of MBP-AT-hook fusion protein and 0.125 μM of labeled DNA and RNA. The AT-hook mutant has two central Arg residues that are changed to Ala. (C) The fraction of DNA (triangle) and RNA (circle) bound by wild type (closed symbol) and mutant (open symbol) fusion protein as determined from (B) is plotted versus concentration of fusion protein. The estimated KD for WT AT-hook binding RNA and DNA are respectively 0.41 and 1.8 μM and for mutant AT-hook RNA and DNA are >>20 μM. Error bars indicate SD from the mean value for three independent replicates. (D) Brg1 binding is determined using CUT&RUN in cells treated with DMSO only (DMSO) or actinomycin D in DMSO (ACT, 3 μg/ml for 4 hr) or triptolide in DMSO (TRP, 10 μM for 2 hr). Brg1 peaks are divided by whether Brg1 is lost (upper panel) or retained (lower panel) in naïve (left) or primed (right) stages. Heatmaps are sorted based on DMSO Brg1 and N represents total number of analyzed peaks. (E) The Brg1 peaks observed with or without DMSO as in Figure S2B are used to determine the overlap between Brg1 peaks lost with addition of actinomycin D or triptolide versus Brg1 peaks lost when the AT-hook is deleted for naïve (left) and primed (right) stages.

Figure 3.. The AT-hook of Brg1 associates in cis with eRNA in the primed stage.

(A-B) The RNA peaks shown are only detected when wild type Brg1 (WT) is pull down and not when the AT-hook of Brg1 is deleted in two independent clones (ΔAT1 and ΔAT2) or Brg1 is not tagged with hemagglutinin (no-HA). (C-D) Some RNA peaks as shown are also detected either in (C) both wild and AT-hook deletion mutant Brg1 or (D) only with the AT-hook deletion mutant Brg1. All of the RNA in (A-D) are only from the intergenic and intronic regions. (E) Shown are the number of RIP-seq peaks within a specified distance from Brg1 CUT&RUN peaks. The distances start from 1 bp to 1 kb, 1 kb to 5 kb and continue with that pattern. (F) The number of Brg1 CUT&RUN and RIP-seq peaks within 1 kb of each other in the primed stage is shown. Heatmaps are sorted based on RIP-seq peaks from wild type Brg1, sorted from highest to lowest and N represents number of peaks. The numbers within the parenthesis correspond respectively to the number of Brg1 CUT&RUN and RIP-seq peaks that overlap.

Figure 4.. Deletion of Brg1’s AT-hook uncouples Brg1 recruitment from eRNA transcription.

(A-B) Meta-analysis of enhancer RNA (eRNA) detected using PRO-seq shows eRNA is transcribed in a stage-specific manner at the sites where Brg1 is recruited specifically in either the (A) naïve or (B) primed stage to intronic and intergenic regions. (C-F) Meta-analysis shows the effect of deleting the AT-hook on (C-D) eRNA transcription and (E-F) Brg1 recruitment in two independent clones (ΔAT1 and ΔAT2) for (C,E) naïve and (D,F) primed stage specific sites in (A-B). (G-H) The peak profiles for Brg1 CUT&RUN and PRO-seq are shown for loci representative of a naïve and primed specific sites where Brg1 recruitment is dependent on the AT-hook like that shown in (E-F).

Figure 5.. Brg1 and its AT-hook are required for recruitment of co-activators MLL3/4 and Med1 and acetylation of lysine 27 of histone H3 (H3K27ac) at enhancers.

(A) Genomic distributions are shown for MLL3/4, Med1, and H3K27ac in the naïve and primed stages as detected by CUT&RUN. (B) The localization pattern of MLL3/4, Med1, and H3K27ac are shown for the naïve and primed stage and clustered into naive only (top), primed only (middle) and both (bottom). (C-E) The impact of deleting the AT-hook of Brg1 on the primed stage specific recruitment/localization of (C) MLL3/4, (D) Med1 and (E) H3K27ac is shown in red, similar to (B). The localization of Brg1 at these sites is also shown in blue. Heatmaps are sorted based on wild type MLL3/4, Med1 of H3K27ac and N represents the number of peaks. (F) The number of peaks overlapping between Med1, MLL3/4, and H3K27ac are shown for sites with high Brg1 and RNAPII in the primed and naive stages.

Figure 6.. Deletion of Brg1’s AT-hook does not change eRNA synthesis or RNAPII localization at sites where MLL3/4, Med1, and H3K27ac are located only in the primed stage.

(A-C) The meta-analysis of RNAPII is shown for sites with high (black) and low (grey) Brg1 occupancy at primed specific peaks of (A) MLL3/4, (B) Med1, and (C) H3K27ac. (D-I) The meta-analysis for (D-H) RNAPII (CUT&RUN) and (G-I) eRNA (PRO-seq) are shown for primed specific peaks of (D,G) MLL3/4, (E,H) Med1 and (F,I) H3K27ac.

Figure 7.. Brg1’s AT-hook is required for activation of stage-specific typical enhancers (TEs) and super-enhancers (SEs) in naive and primed stages

(A-C) Distinct classes of super-enhancers in the primed stage are identified based on the localization patterns of (A) Med1, (B) MLL3/4, or (C) H3K27ac using the ROSE algorithm. (D) The number of active enhancer peaks from (5F) located within the SEs (green) identified in (A-C) are shown plus the number of peaks that are not (blue), representative of TEs. (E) The total number of SEs with co-activator peaks associated with highly (red) or lowly enriched (black) Brg1 and RNAPII are shown for each co-activator in naïve or primed stages. (F-G) Shown are the genes down regulated by deletion of the AT-hook within the indicated distances from AT-hook dependent stage-specific TEs (F) and SEs (G) in naive (black) and primed (grey) stages. Distances are shown for intervals starting with 1–10 kb progressing towards 751 kb to 1 Mb. Line plots show the cumulative frequency of the differential expressed genes in naïve (blue) and primed (red) states. (H-I) Meta-analysis is shown for nascent transcription of the genes within 100 kb of TEs and/or 1Mb of SEs that are (H) primed or (I) naïve stage specific.