Long-range regulation of transcription scales with genomic distance in a gene-specific manner

Abstract

Although critical for tuning the timing and level of transcription, enhancer communication with distal promoters is not well understood. Here, we bypass the need for sequence-specific transcription factors (TFs) and recruit activators directly using a chimeric array of gRNA oligos to target dCas9 fused to the activator VP64-p65-Rta (CARGO-VPR). We show that this approach achieves effective activator recruitment to arbitrary genomic sites, even those inaccessible when targeted with a single guide. We utilize CARGO-VPR across the Prdm8-Fgf5 locus in mouse embryonic stem cells (mESCs), where neither gene is expressed. Although activator recruitment to any tested region results in the transcriptional induction of at least one gene, the expression level strongly depends on the genomic distance between the promoter and activator recruitment site. However, the expression-distance relationship for each gene scales distinctly in a manner not attributable to differences in 3D contact frequency, promoter DNA sequence, or the presence of repressive chromatin marks at the locus.

Keywords: CRISPR activation; dCas9; enhancer; gene regulation; genomic distance; promoter.

Figure 1.. Lack of dCas9 binding confounds interpretation of CRISPRa screen

A. Schematic demonstrating two scenarios in which long-range activation is limited to privileged sites (upper) or is compatible with any site given sufficient cofactor recruitment (lower). B. Genome browser tracks of CTCF (ChIP-seq), CTCF motif orientations (+ strand, red, - strand, blue), H3K27ac (ChIP-seq) and ATAC-seq at the Prdm8-Fgf5 locus in mESCs (top three tracks) and EpiLCs (bottom two tracks). TAD position is highlighted as grey bar at the top. C. Schematic of CRISPRa screen to identify regions capable of activating Fgf5-eGFP. D. Enrichment of guides in the eGFP+ population relative to the unsorted population (over 13,000 guides screened, lower black bars). E. Schematic of possible interpretations of negative CRISPRa results F. Positions of single guides tested for dCas9 binding. Gold bars indicate single guides unable to activate Fgf5 in CRIPRa screen. Green bars are activating guides (enriched in the eGFP+ population). G. RT-qPCR of Fgf5 expression in cells expressing dCas9-VPR and select single guides, as indicated in F. H. dCas9 ChIP-qPCR from cells expressing dCas9-VPR and select single guides (same guides as F and G).

Figure 2.. CARGO-VPR enables dCas9 recruitment to previously inaccessible sites

A. Schematic of CARGO-VPR strategy, utilizing a 6-mer gRNA arrays (CARGO arrays) for dCas9-VPR targeting to genomic regions of interest B. Positions of twelve individual guides, two 6-mer arrays, and one 12-mer array designed to target a region 14kb upstream of the Fgf5 promoter (14 kb USF). Locations of PCR amplicons used in dCas9 ChIP-qPCR are shown as colored lines labeled A, B, and C. C. dCas9 ChIP-qPCR percent input recovery at sites labeled in (B) in cells expressing dCas9-VPR and single guides (non-targeting (NT), 3,5, or 10), 6-mer arrays (1–6 or 7–12), or a 12-mer array (1–12). Colored dots are the mean of two biological replicates and grey dots are the individual replicates. D. RT-qPCR of Fgf5 expression in cells expressing dCas9-VPR and single guides (NT, 3,5, or 10), 6-mer arrays (1–6 or 7–12), or a 12-mer array (1–12). Colored dots are the mean of three biological replicates and grey dots are the individual replicates. E. Positions of 6-mer CARGO arrays at the Prdm8-Fgf5 locus. F. dCas9, G. RNAPII, and H. H3K27ac ChIP-qPCR in cells expressing CARGO arrays shown in (E) or a NT guide. Individual panels correspond to a PCR amplicon overlapping indicated array. Bar height is the mean of n=3–5 replicates with each replicate shown as black dots (*=p<0.05,**=p<0.01, ***p<0.001, Welch’s two sample t-test). I. RT-qPCR of Prdm8 (upper) and Fgf5 (lower) in cells expressing indicated CARGO-VPR target. Expression shown as log₁₀ ratio of normalized expression in the CARGO-VPR line over normalized expression in the NT control. Colored points are the mean of 4 replicates. Grey points are individual expression values. J. Expression normalized by max observed for each CARGO-VPR target plotted against percent input recovery in cells expressing dCas9-VPR and 1, 3, 4, 5, or 6 guides. K. negative log p-values for the binding coefficient in a linear regression model of expression as a function binding (percent input recovery) for models including binding data at different, indicated percent input recovery thresholds. See also Figure S1

Figure 3.. Gene-specific decay profiles of expression with genomic distance

A. Positions of 38 CARGO-VPR arrays targeting the Prdm8-Fgf5 locus. B. Schematic of CARGO-VPR line generation. After dCas9-VPR clonal lines are isolated, each CARGO array is stably integrated using PhiC integrase generating 38 cell populations. Each population is treated with doxycycline for three days prior to downstream analyses such as RT-qPCR and ChIP-qPCR. C. dCas9 ChIP-qPCR in all 38 CARGO-VPR cell populations. Black points are the mean of n= 3–4 biological replicates. Error bars are standard deviation. Grey points indicate the same qPCR amplicon assayed from a cell population expressing a non-targeting guide (n=2). D. RT-qPCR of Prdm8 (upper, purple) and Fgf5 (lower, green) from cell populations expressing dCas9-VPR and a CARGO array targeted to each of the positions shown on x-axis. Colored dots are the mean of at least four replicates. Grey points are individual expression values. Expression is plotted as expression normalized to a housekeeping gene, Rpl13a, over normalized expression from a population expressing a non-targeting guide. E. RT-qPCR of Prdm8 (Left) and Fgf5 (Right) plotted against absolute distance of array from the target promoter. Data points from arrays located upstream or downstream from the promoter are shown as triangles and dots, respectively. Colored dots/triangles are the mean of at least four replicates. Grey points are individual expression values. Expression is plotted as expression normalized to housekeeping gene, Rpl13a, over normalized expression from a population expression a non-targeting guide. F. Log₁₀ RT-qPCR expression in CARGO-VPR lines normalized to expression in non-targeting line plotted against log₁₀ distance of CARGO-VPR recruitment site and promoter. Arrays that have an average percent input recovery less than 3.5% were excluded from analysis. Spline regression fits with slopes of −1.77 and −0.18 for Prdm8 and −0.34 and −4.30 for Fgf5 (adjusted R² 0.94, 0.87,respectively). See also Figure S2

Figure 4.. 3D contact frequencies do not explain distinct expression profiles

A. Map of 38 5kb ORCA probes in relation to Prdm8 and Fgf5 and selected CARGO arrays that activate Prdm8 (Purple, 7.5kb USP) or Fgf5 (Green, 33kb USF). B. Schematic of ORCA strategy C. Pairwise median distance plots from the probes overlapping the Prdm8 promoter or the Fgf5 promoter (black lines) and all other probes. Location of individual probes is shown as rainbow-colored blocks and corresponds to placement in (A). Purple and green shading represent 95% confidence interval of the median. D. Median pairwise distance from probe overlapping Prdm8 promoter (upper) or Fgf5 promoter (lower) and other probes in cells expressing a non-targeting guide compared with equivalent median pairwise distances in cells expressing CARGO arrays in (D). Dotted lines are y=x. E. Distribution of pairwise distances between probes overlapping the Prdm8 promoter and the 7.5kb USP CARGO-VPR site in dCas9-VPR cells expressing a non-targeting guide (NT) or 7.5kb USP CARGO (7.5kb USP). F. Distribution of pairwise distances between probes overlapping the Fgf5 promoter and the 33kb USF CARGO-VPR site in dCas9-VPR cells expressing a non-targeting guide (NT) or 33kb USP CARGO (33b USF) G. Expression normalized to max observed of Prdm8 (upper) and Fgf5 (lower) induced by CARGO-VPR arrays with strong binding (>3.5%) (y-axis) by contact frequency (x-axis) between the promoter and the CARGO target site, as extrapolated from ORCA measurements. Colored dots are the mean of at least four replicates. Grey points are individual expression values. See also Figure S3

Figure 5.. Promoter swap does not alter scaling of expression responses

A. Schematic of the Prdm8 and Fgf5 promoter swap. 450bp sequence surrounding each TSS was swapped in P2FF2P mESCs (see Methods for details). B. RT-qPCR expression of Fgf5 and Prdm8 in WT and promoter swap (P2FF2P) lines expressing dCas9-VPR and CARGO arrays. Colored points are the mean of three replicates. Grey points are individual expression values. C. Comparison of Fgf5 and Prdm8 expression levels in promoter swap (P2FF2P) vs WT mESCs upon CARGO VPR. line is y=x. Points are mean of three biological replicates. See also Figure S4