Human senataxin resolves RNA/DNA hybrids formed at transcriptional pause sites to promote Xrn2-dependent termination

Abstract

We present a molecular dissection of pause site-dependent transcriptional termination for mammalian RNA polymerase II (Pol II)-transcribed genes. We show that nascent transcripts form RNA/DNA hybrid structures (R-loops) behind elongating Pol II and are especially prevalent over G-rich pause sites positioned downstream of gene poly(A) signals. Senataxin, a helicase protein associated with AOA2/ALS4 neurodegenerative disorders, acts to resolve these R-loop structures and by so doing allows access of the 5'-3' exonuclease Xrn2 at 3' cleavage poly(A) sites. This affords 3' transcript degradation and consequent Pol II termination. In effect, R-loops formed over G-rich pause sites, followed by their resolution by senataxin, are key steps in the termination process.

Graphical abstract

Figure 1

Transcriptional Termination Defect of Constructs with G-Rich Pause Sites in Senataxin-Deficient HeLa Cells (A) RNAi-mediated depletion of human senataxin in HeLa cells. Left panel: qRT-PCR analysis of mock-treated and senataxin siRNA1-treated HeLa cells. Bars represent average values ± SD from three independent biological experiments. 5S rRNA was used as a control. The amount of SETX mRNA in mock cells was taken as 100%. Right panel: western blot analysis of 60 μg of protein extracts from mock-treated and senataxin siRNA1-treated HeLa cells using anti-senataxin antibody. E1B-AP5 protein (∼120 kDa) was used as a loading control. Cut-out middle lane of this gel contained lower amount of protein extract. (B) Left panel: diagram of β-globin/β-actin and β-globin/MAZ4 gene constructs, with exons shown as boxes and pause elements indicated. Poly(A) signal is denoted by arrows. P/U3 probe was used in RNase protection analysis and is shown as dashed lines. “A” and “B” in diagram denote positions of PCR products used to calculate the amount of readthrough transcript in (C). Middle panel: RNase protection analysis of β-globin/β-actin and β-globin/MAZ4 constructs from mock-treated and senataxin siRNA1-treated cells performed with P/U3 riboprobe. 85 nt fragment, “P,” derives from transcripts initiated on the HIV promoter, while an ∼240 nt band (U3) derives from readthrough transcription. ^∗ denotes partial RNase digestion product detected in the cytoplasm in some experiments. Right panel: readthrough transcript level was calculated as a signal ratio of U3/P (%), based on three independent experiments; values are ±SD. (C) Termination efficiency of β-globin/β-actin and β-globin/MAZ4 constructs from mock-treated and senataxin siRNA1-treated cells using qRT-PCR analysis on the total RNA. Readthrough transcript level was calculated as a ratio of product A (readthrough RNA) versus product B (gene transcript) from three independent experiments; values are ±SD. The A/B ratio in mock-treated cells was taken as 1. ^∗ indicates statistical significance (p < 0.05), based on unpaired, two-tailed distribution Student's t test.

Figure 2

R-Loops Form over Transfected Gene Constructs and Depend on Transcription, Functional Poly(A) Signal, and G-Rich Pause Elements (A) Top panels: diagram of β-globin/β-actin (left) and β-globin/MAZ4 (right) gene constructs, with positions of primers used in DIP analysis indicated. Bottom panels: DIP analysis of β-globin/β-actin (left) and β-globin/MAZ4 (right) constructs. DIP signal was normalized to exon 1 signal. (B) R-loops are dependent on active transcription. Left panel: R-loop profile of cells transfected with β-globin/β-actin gene construct ± trans-activator TAT. Right panel: RT-PCR analysis of the spliced mRNA detected from HeLa cells transfected with β-globin/β-actin construct ± TAT. “M” indicates the molecular weight DNA markers. DIP signal was normalized to exon 1 signal of β-globin/β-actin construct cotrasfected with TAT. (C) R-loop profile of β-globin/MAZ4 (gray bars), mMAZ4 (black bars), and MAZ4ΔpA (white bars) gene constructs. DIP signal was normalized to exon 1 signal of β-globin/MAZ4 construct. (D) R-loop profile of β-globin/β-actin (left panel) and β-globin/MAZ4 (right panel) gene constructs in mock-treated (gray bars) and senataxin siRNA1-treated (white bars) cells. DIP signal was normalized to exon 1 signal of β-globin/β-actin (left panel) and β-globin/MAZ4 (right panel) constructs in mock-treated cells. DIP profile is based on average values ± SD from at least three independent biological experiments. ^∗ indicates statistical significance (p < 0.05), based on unpaired, two-tailed distribution Student's t test.

Figure 3

RNase H1 Resolves R-Loops and Promotes Transcriptional Readthrough (A) Western blot analysis of 30 μg of protein extracts obtained from HeLa cells transfected with increasing amounts (0–2 μg) of GFP-RNase H1 expression plasmid. Western blot was analyzed using anti-GFP antibody. Actin was used as loading control. (B) Amount of readthrough transcript detected in β-globin/β-actin, β-globin/MAZ4, β-globin/mMAZ4, and ΔCoTC constructs in mock HeLa cells (sample “-”) and HeLa cells, cotransfected with increasing amounts of GFP-RNase H1 plasmid. Readthrough transcript level was calculated as a ratio of A/B, using qRT-PCR analysis (see Figure 1B for primer positions). The A/B ratio in mock-treated cells (sample “-”) was taken as 1. This data is also presented as values for transcripts “A” and “B” separately, normalized to 5S rRNA (Figure S2). (C) Top panel: diagram of β-globin/MAZ4 gene construct with positions of NRO probes indicated. Bottom left and right panels: NRO analysis of β-globin/MAZ4 construct in mock-treated cells (black bars on the right panel), HeLa cells cotransfected with 2 μg of GFP-RNase H1 (gray bars), and senataxin siRNA1-treated cells (white bars). M13 represents a background probe. Right panel: NRO signals were quantified by PhosphoImager analysis. Background (signal from probe M13) was subtracted from each hybridization signal normalized to probe B3, taken as 1. All bars represent average values ± SD from three independent biological experiments. (D) DIP analysis of β-globin/MAZ4 construct following overexpression of 2 μg of GFP-RNase H1 in HeLa cells. Positions of primers used in DIP analysis are shown in Figure 2A. DIP profiles are based on average values ± SD from three independent biological experiments. DIP signal was normalized to exon 1 signal of β-globin/MAZ4 construct without GFP-RNase H1 overexpression. (E) Amount of readthrough β-globin/MAZ4 transcript in mock-treated and senataxin-depleted HeLa cells with or without overexpression of 2 μg of GFP-RNase H1 plasmid. Readthrough transcript level was calculated as a ratio of A/B, using qRT-PCR analysis (for primer positions see Figure 1B). The A/B ratio in mock-treated cells (−RNase H1 sample) was taken as 1. Average qRT-PCR values ± SD from four independent biological experiments are presented. ^∗ indicates statistical significance (p < 0.05), based on unpaired, two-tailed distribution Student's t test.

Figure 4

Senataxin Is Required for Pol II Transcriptional Initiation, Termination, and Xrn2 Recruitment to the Termination Regions of the Endogenous β-Actin Gene (A) Diagram of β-actin gene, with exons shown as white boxes and pause element shown as gray box. Positions of primers used in ChIP analysis are shown above the diagram. ChIP analysis was carried out with Pol II (left panel) or senataxin (right panel) antibody. (B) Pol II ChIP analysis of β-actin gene in mock-treated (gray bars) and senataxin siRNA1-treated (white bars) cells. (C) Br-UTP NRO analysis of β-actin gene in mock-treated (gray bars) and senataxin siRNA1-treated (white bars) cells. Amount of nascent Br-UTP RNA was normalized to intron 3 probe in mock-treated cells. All bars represent average values ± SD from at least three independent biological experiments. Enrichment of readthrough transcripts for probes “5′ pause,” “pause,” “C,” and “D” in senataxin-depleted cells, as compared to mock cells, was calculated relative to intron 1 and intron 3 signals. (D) Top panel: DIP analysis of β-actin gene using RNA/DNA hybrids-specific S9.6 antibody, based on average values ± SD from at least three independent biological experiments. DIP signal was normalized to intron 1 signal. Bottom panel: R-loops formed over β-actin gene are sensitive to RNase H and only partially to S1 nuclease digestion. DIP samples were either untreated (gray) or treated with 10 U of RNase H (black bars) or 200 U of S1 nuclease (white bars). (E) Xrn2 ChIP of β-actin in mock-treated (gray bars) and senataxin-depleted (white bars) HeLa cells. ChIP values are based on average values ± SD from at least three independent biological experiments. ^∗ indicates statistical significance (p < 0.05), based on unpaired, two-tailed distribution Student's t-test.

Figure 5

Model for Role of Senataxin and R-Loops in Transcriptional Termination DNA is shown as solid lines and RNA as a dotted line. R-loops are formed between the ssDNA template strand and nascent RNA behind the elongating Pol II. Vertical black lines denote RNA/DNA hybrids. Senataxin is shown as a gray oval over the R-loop region. R-loops formed over the G-rich pause region of human β-actin gene are necessary for Pol II to pause downstream of the poly(A) site. Senataxin is needed to resolve R-loop structures and so allow 5′–3′ exonuclease Xrn2 to degrade the nascent RNA from the site of poly(A) cleavage and catch up with paused Pol II, causing its transcriptional termination.