Figure 1.
Sequence analysis of the FMR1 promoter reveals signatures of R-loop formation.
GC skew (red, left y-axis), CpG observed/expected ratio (CpG O/E; navy, right y-axis), and GC% (gray, right y-axis) calculated over a sliding 100 nt window from −500 to +1200 nt around the downstream-most known transcription start site (vertical dotted line). Gray-shaded box highlights CGI defined by CpG O/E>0.6 (navy dotted lines) and GC%>50% for at least 200 nt. Schematic at the top shows the FMR1 5′UTR with multiple transcription start sites (black arrows), G-clusters (red ticks), and CGG repeats (striped box), all overlapping the CGI (gray bar) for scale to the graph below.
Figure 2.
R-loop pull-down in human dermal fibroblasts confirms R-loop formation in the genome.
(A) Fold enrichment for FMR1 in dermal fibroblast cells cultured from seven individuals using a monoclonal antibody specific to hybrids. Enrichment is relative to input and a non-R-loop-forming genomic reference locus. (B) Treatment with recombinant RNases H1 and H2 (RNase H) eliminates enrichment seen for FMR1 (solid lines) and MYADM (broken lines).
Figure 3.
Effect of transcription and repeat length on FMR1 R-loop formation.
(A) Schematic of DOX-ON constructs with short or expanded FMR1 CGG repeats or non-FMR1 sequence, each with GFP reporter tags. Black arrowheads mark sites of restriction enzyme cleavage prior to DRIP, with EcoRI cutting at the start of the FMR1 5′UTR and XbaI cutting at the end of EGFP. (B) mRNA expression relative to non-induced cells for each construct. Error bars: SEM from 2 biological replicates. (C) DRIP fold enrichment of GFP fragment relative to the episome backbone. Error bars: SEM from 3 biological replicates. (D) DRIP percentage of input normalized to peak recovery (6 hours DOX ON) of GFP fragment at 0, 1, 2, and 24 hours post DOX washout, and No-DOX treatment. Error bars: SEM from 3 biological replicates.
Figure 4.
Non-denaturing bisulfite footprinting of the displaced DNA strand of the FMR1 R-loop.
Each row represents an individual sequence clone, grouped together for each allele size, from cultured human dermal fibroblasts. Each column is a cytosine position, with filled boxes representing converted, single-stranded DNA and open boxes representing unconverted, double-stranded DNA. Empty boxes represent sequence gaps from bacterial deletion or loss of clean sequencing signal. Schematic diagram at the top represents the FMR1 5′UTR with marked TSSs (black arrows), translation start (ATG), CGG repeats (striped box with orange border), PCR primers (blue arrows), and G-clusters (red ticks; red dotted lines).
Figure 5.
Model of proposed CGG-repeat effects on the FMR1 R-loop.
R-loops that span the FMR1 CGG-repeat region (yellow) during transcription could adopt a hairpin structure within the displaced CGG-repeat strand, thus protecting the CGG-repeat region from bisulfite conversion while leaving both 5′ and 3′ flanking regions exposed; the CGG-repeat is known to form such structures readily in vitro [61]. An alternative structure, although less energetically feasible, would involve maintenance of R-loops flanking the CGG-repeat element, which has collapsed into a dsDNA structure again. Loss of the upstream R-loop region would explain the absence of bisulfite conversion in ∼25–50% of molecules (Figure 4). Red, nascent RNA transcript; 90° arrow, start of transcription; blue sphere, Pol II.