Senataxin Mutation Reveals How R-Loops Promote Transcription by Blocking DNA Methylation at Gene Promoters

Abstract

R-loops are three-stranded nucleic acid structures found abundantly and yet often viewed as by-products of transcription. Studying cells from patients with a motor neuron disease (amyotrophic lateral sclerosis 4 [ALS4]) caused by a mutation in senataxin, we uncovered how R-loops promote transcription. In ALS4 patients, the senataxin mutation depletes R-loops with a consequent effect on gene expression. With fewer R-loops in ALS4 cells, the expression of BAMBI, a negative regulator of transforming growth factor β (TGF-β), is reduced; that then leads to the activation of the TGF-β pathway. We uncovered that genome-wide R-loops influence promoter methylation of over 1,200 human genes. DNA methyl-transferase 1 favors binding to double-stranded DNA over R-loops. Thus, in forming R-loops, nascent RNA blocks DNA methylation and promotes further transcription. Hence, our results show that nucleic acid structures, in addition to sequences, influence the binding and activity of regulatory proteins.

Keywords: ALS; ALS4; DNA methylation; R-loop; TGFB; amyotrophic lateral sclerosis; helicase; motor neuron disease; senataxin; transcription.

Figure 1. Clinical features of 18 ALS4 subjects

(A) Age of onset is defined as the age of initial muscle weakness. All patients had electromyography (EMG) showing denervation in at least two different muscle groups. Hyperreflexia was scored as increased reflexes in either the upper or lower extremities; diminution of reflexes was observed in patients with severe weakness. Strength was evaluated using the Medical Research Council (MRC) scale (https://www.mrc.ac.uk/research/facilities-and-resources-for-researchers/mrc-scales/mrc-muscle-scale/) with weakness determined by MRC scale less than 5 in any muscle group. Falls were assessed retrospectively; (+) = at least one fall in the past year. LE = lower extremity, UE = upper extremity. Dysmetria identified as positive findings on finger-to-nose or past pointing tests. Creatine kinase and alpha-fetoprotein levels were measured in 15 patients (average age 41) compared to 8 related, unaffected controls in parentheses (average age 50); values shown are mean ± SEM). (B) T1 MRI measurement of thigh muscle volume. Muscle volume correlates significantly with disease duration (defined as years since diagnosis) (R²=0.54, P=0.02).

Figure 2. Fewer R-loops in ALS4 patients

(A) Dot-blot probed with S9.6 antibody shows fewer R-loops in ALS4 fibroblasts compared to control. A duplicate blot was probed with anti-dsDNA antibody as loading control. S9.6 intensity was normalized to that of the dsDNA and average was shown. (n= 3 control & 3 patients; error bars are SEM). * represents P<0.05 (t-test). (B) Immunofluorescence staining of ALS4 fibroblasts shows a significant reduction in nuclear S9.6 signal that co-localizes with DAPI (scale bar = 7.5μm) as compared to controls. n = 3 controls & 3 patients, 10 fields of view were counted per sample and quantified using ImageJ. Arrows indicate nuclear R-loops. R-loop staining with the S9.6 antibody shows a significant reduction of R-loop signal in cell nuclei in the (C) lumbar spinal cord and (D) the motor cortex. NeuN is shown as a control of neuronal nuclei staining. Neuronal nuclei indicated by arrows. n = 2 controls & 2 patients; scale bar = 25 m, 10 fields per sample, error bars are SEM. *** represents P<0.001 (t-test). (E) Fewer R-loops in cells transfected with L389S form of SETX compared to WT (t-test, P=0.06). R-loops were analyzed by dot blot using S9.6 antibody. Error bars are SEM of triplicates. S9.6 signal was first normalized to loading control (dsDNA) and then to SETX expression as detected by western blot. (F) Immunofluorescence staining showing reduced nuclear S9.6 signal (green) in cells transfected with L389S form of SETX compared to WT. Note in both L389S and WT SETX transfected cells, S9.6 staining is reduced in those cells with higher SETX expression (red); scale bar = 5 μm, merged image with staining of S9.6, SETX, and DAPI (blue), R-loops in nuclei indicated by arrows.

Figure 3. TGFβ is activated in ALS4 patients

(A) Scatter plot of gene expression levels from RNA-seq of fibroblasts (n = 3 controls & 3 ALS4 patients); differentially expressed genes that point to activation of TGFβ pathway in the ALS4 patients are shown in orange; genes shown are those with ≥50% differences. (B) Expression levels of genes (RT-PCR) in TGFβ pathway from fibroblasts (n = 5 controls & 5 ALS4 patients; 2 replicates from each sample), error bars are SEM. Genes shown are those with ≥20% differences. (C) & (D) Increased staining of BMP2 and pSMAD2/3 in patient lumbar spinal cords; n = 2 controls & 2 ALS4 patients; scale bar = 25μm, 10 fields per sample, error bars are SEM. ** represents P<0.01, *** represents P<0.001 (t-test).

Figure 4. Fewer promoter R-loops and lower expression of BAMBI in ALS4

(A) BAMBI promoter is GC rich and G-skewed in comparison to median levels of all genes annotated in RefSeq. 50-bp sliding windows were used to compute GC% = [(G+C)/(G+C+A+T)]%, and GC skew = (G−C)/(G+C). The GC% and GC skew at the BAMBI promoter (in red), and their median level for all 23,682 promoters (in green) are shown. Shaded area represents the range between 40^th and 60^th percentile. (B) S9.6 DNA-RNA IP followed by quantitative PCR with primers specific to a region containing known R-loops at β-actin or BAMBI promoters (n = 2 controls and 2 patients). Primers specific to intronic regions free of R-loops were used as negative controls. Arrows indicate location of primers. (C) S9.6 DNA-RNA IP followed by sequencing in control and ALS4 cells. Normalized read counts across BAMBI gene are plotted. Transcription start site (TSS), Cleavage and polyadenylation site (CPS). (D) Overexpression of human RNase H1 in fibroblasts led to fewer R-loops β-actin or BAMBI promoters (same PCR primers used in 4B), and lower BAMBI transcript (E) and protein (F) expressions. 8 fields of view with 15-20 cells per field were quantified. P<0.001, scale bar = 15μm. (G) A 72-hour time course showed BAMBI gene expression correlates negatively with the expression of RNase H1. (H) Overexpression of L389S form of senataxin led to lower BAMBI gene expression (t-test, P<0.01). Error bars are SEM.

Figure 5. R-loops regulate BAMBI expression by reducing promoter methylation

(A) Cytosine methylation is higher in ALS4 patients than in controls. From methyl-C sequencing reads, at each genomic position, level of cytosine methylation was calculated as percentage of methylated C. Upper panel: average levels of methylated cytosine in fibroblasts at BAMBI promoter are shown (n = 5 controls & 5 ALS4 patients). Lower panel: heatmap showing methylated cytosine levels from each subject. Each row shows the average methyl-C level across the BAMBI promoter region (TSS±500bp). (B) DNMT1 ChIP-qPCR shows significantly more DNMT1 binds to BAMBI promoter in ALS4 patient compared to control (P<0.05; error bar: SEM of quadruplicates.). Arrow indicates location of primers. (C) DNMT1 knockdown results in lower DNMT1 expression and higher BAMBI expression; error bars represent SEM of triplicates. (D) RNase H1 overexpression which resolves R-loops leads to more DNMT1 binding (top) and consequently higher methylation (bottom) of BAMBI promoter. (P<0.05, error bars represent SEM of triplicates). Average methylation levels of cytosines within BAMBI promoter are shown. (E) R-loop-mediated repression of BAMBI expression is dependent on DNMT1. BAMBI expression did not change following RNase H1 overexpression when DNMT1 was knocked down (error bars represent SEM of triplicates).

Figure 6. DNMT1 favors binding and methylation of dsDNA over RNA/DNA hybrid

(A) DNMT1 activity is higher with dsDNA than RNA-DNA hybrid as substrates; error bars represent SEM of triplicates. (B) Binding affinity of DNMT1 for dsDNA is higher than for RNA/DNA hybrid. EMSA was used to measure binding of DNMT1 to BAMBI promoter represented as biotinylated dsDNA or corresponding RNA/DNA hybrid. Unlabeled dsDNA was added as competitor to show specificity of binding. (C) More binding of DNMT1 to dsDNA than RNA/DNA hybrid. Binding with a dilution series of DNMT1 to 2.5nM dsDNA (solid lines) or RNA/DNA hybrid (dotted lines) assayed by biolayer interferometry. Baseline was recorded from 0 to 300 seconds, association of DNMT1 with dsDNA or RNA/DNA hybrid from 300-900 seconds, followed by dissociation. (D) ALS4 patients have less nascent transcripts of BAMBI. In nuclear run-on (NRO) of primary fibroblasts (n = 2 controls & 2 ALS4 patients), nascent transcript level was measured by quantitative PCR using primers specific to promoter of BAMBI, and normalized to promoter of a housekeeping gene ABLI that has no R-loops in S9.6 pull-down assay.

Figure 7. R-loops regulate promoter methylation of more than 1,200 human genes

(A) -1,386 gene promoters that are GC-rich (GC%>0.6) and have R-loops. Metagene plot of DRIP-Seq data shows enrichment of R-loops near promoters (top). Metagene plots of DRIP-seq also show genome-wide decrease of R-loops in ALS4 (bottom) compared to the control (top). Average ± 95% CI was plotted. (B) Among the 1, 386 genes with GC-rich promoters, 1,240 genes showed an increase in DNA methylation following RNaseH1 overexpression; fold changes at two time points are shown. (C) Among the genes that showed R-loop dependent methylation in (B), 697 gene showed changes in gene expression following overexpression of RNase H1, of which 508 showed decreased expression.