Enhancer Regulation of Transcriptional Bursting Parameters Revealed by Forced Chromatin Looping

Abstract

Mammalian genes transcribe RNA not continuously, but in bursts. Transcriptional output can be modulated by altering burst fraction or burst size, but how regulatory elements control bursting parameters remains unclear. Single-molecule RNA FISH experiments revealed that the β-globin enhancer (LCR) predominantly augments transcriptional burst fraction of the β-globin gene with modest stimulation of burst size. To specifically measure the impact of long-range chromatin contacts on transcriptional bursting, we forced an LCR-β-globin promoter chromatin loop. We observed that raising contact frequencies increases burst fraction but not burst size. In cells in which two developmentally distinct LCR-regulated globin genes are cotranscribed in cis, burst sizes of both genes are comparable. However, allelic co-transcription of both genes is statistically disfavored, suggesting mutually exclusive LCR-gene contacts. These results are consistent with competition between the β-type globin genes for LCR contacts and suggest that LCR-promoter loops are formed and released with rapid kinetics.

Fig 1. β-globin transcriptional burst fraction and size increase during erythroid maturation

A) Model of how bursts can be modulated during erythroid maturation to increase β-globin mRNA levels. B) Single-molecule RNA FISH of β-globin intron and exon to identify transcription sites in G1E-ER4 cells, 4 or 24 hours after estradiol addition (arrows indicate transcription sites, intron and exon channels are set to same intensities across images, maximum merges [maximum signal from 45 stacks] shown). C) Representative experiment showing fluorescence intensities of β-globin transcription sites in G1E-ER4 cells at indicated time points after estradiol addition (tics on x axis=medians for this experiment). D) Mean number of alleles transcribing β-globin per cell at time points after estradiol addition. N=3 biological replicates. E) Median fluorescence intensities of β-globin transcription sites in G1E-ER4 cells following estradiol addition. N=3 biological replicates. F) Mean anti-RNA polymerase II ChIP in G1E-ER4 cells at indicated times after estradiol addition in the globin locus normalized to input. Primer pairs targeting distinct regions of the gene are listed on the x-axis. The silent CD4 locus served as a negative control. N=3 biological replicates. G) Mean ratio of RNA polymerase II ChIP signal at β-globin exon 2 to β-globin promoter. N=3 biological replicates. Error bars represent SEM. See also Figures S1 and S2 and Table S1.

Fig 2. Deletion of the LCR predominantly impairs β-globin burst fraction

A) GATA1 ChIP-Seq browser track view of the β-globin locus. Peaks demarcate the β-globin promoter and DNase I hypersensitive sites of the LCR. Guide RNA placements for Cas9-mediated LCR deletions are indicated. B) Single-molecule RNA FISH using β-globin intron and exon probes to identify transcription sites in wild-type and LCR-deleted G1E-ER4 cells 24 hours after estradiol addition (arrows indicate transcription sites, intron channel is set to same intensities across images, maximum merges shown). C) Mean number of alleles transcribing β-globin per cell in G1E-ER4 wild-type and LCR-deleted cells, treated with estradiol for 24 hours. N=3 biological replicates. D) Fluorescence intensity of β-globin transcription sites in G1E-ER4 parental and LCR-deleted cells, exposed to estradiol for 24 hours. Wilcoxon non-parametric t-test, pooling the two LCR-edited clones, p=1.54e-10. Data from 3 biological replicates pooled. Error bars represent SEM. See also Figure S3 and Table S2.

Fig 3. Forcing LCR-promoter contacts increases β-globin burst fraction independent of burst size

A) Schematic of the forced looping strategy in murine G1E cells lacking GATA1. B) Mean number of alleles transcribing β-globin per cell infected with empty vector or vector expressing mZF-SA. N=3 biological replicates. C) Fluorescence intensity of β-globin transcription sites. Wilcoxon non-parametric test comparing control and mZF-SA intensities, p=0.42. Data were pooled from 3 biological replicates. D) Schematic of the forced looping strategy in murine G1E-ER4 cells induced for 9 hours with estradiol. E) Mean number of alleles transcribing β-globin per cell in G1E-ER4 cells infected either with control vector or mZF-SA, 9 hours after estradiol addition. N=3 biological replicates. F) Representative experiment showing fluorescence intensity of β-globin transcription sites in G1E-ER4 cells, infected either with control vector or vector expressing mZF-SA, 9 hours after estradiol addition. G) Median fluorescence intensity of β-globin transcription sites in G1E-ER4 cells infected either with control vector or mZF-SA, 9 hours after estradiol addition. N=3 biological replicates. Error bars represent SEM. See also Figure S4 and Table S3.

Fig 4. Effects of maturation and forced LCR-promoter contacts on transcriptional bursting parameters in human erythroid cells

A) Mean number of alleles transcribing β-globin per cell, and B) median fluorescence intensity of β-globin transcription sites at 0 or 3 days of maturation induction. N=3 human donors. C) Schematic of forced LCR-γ-globin promoter loop strategy in primary human erythroid cells. D) Single-molecule RNA FISH of γ- and β-globin introns in human primary erythroid cells infected with hZF-SA expressing vector or control vector. (γ- and β-globin intron channels set to same intensities across images, maximum merges shown). E) Mean number of alleles transcribing γ-globin per cell in hZF-SA expressing cells or control cells, normalized to proportion transcribing in the control vector condition. Wilcoxon paired t-test, p=0.016. N=5 human donors. F) Mean number of alleles transcribing β-globin per cell normalized to control vector condition. Wilcoxon paired t-test, p=0.42, or p=0.2 excluding outlier. N=5 human donors. G) Median fluorescence intensities of γ-globin transcription sites cells infected with control vector or hZF-SA vector, normalized to proportion transcribing in the control vector condition. Wilcoxon paired t-test, p=0.84. N=5 human donors. H) Median fluorescence intensities of β-globin transcription sites in cells infected with control vector or hZF-SA vector, normalized to proportion transcribing in the control vector condition. Wilcoxon paired t-test, p=0.09. N=5 human donors. Error bars represent SEM. See also Figure S4 and Table S4.

Figure 5. β- and γ-globin promoters compete for enhancer activity when positioned in cis

A) Representative image of a primary human erythroid cell co-transcribing β- and γ-globin from the same allele. (Arrowhead indicates co-transcribed transcription site, maximum merges shown.) B) Model of how allelic β- and γ-globin transcription was quantitated using RNA FISH data. C) Example quantification of cellular β- and γ-globin transcription for one donor. Fisher exact test odds ratio=0.66, p-value=0.13. D) Fisher exact test odds ratio and 95% confidence interval for 4 human samples for cis-competition of globin alleles, estimating 3.2 alleles per cell. E) Median fluorescence intensity of β-globin transcription sites in primary human cells, either non-colocalized transcription sites or those colocalized with γ-globin transcription sites. p=0.4, Wilcoxon paired rank-sum t-test, N=4 donors. Error bars represent SEM. F) Fluorescence intensity of β- and γ-globin transcription sites for each co-transcriptional transcription site, Pearson R²=0.003, N=4 donors. See also Figure S5 and Table S5.