Human capping enzyme promotes formation of transcriptional R loops in vitro

Abstract

Cap formation is the first step of pre-mRNA processing in eukaryotic cells. Immediately after transcription initiation, capping enzyme (CE) is recruited to RNA polymerase II (Pol II) by the phosphorylated carboxyl-terminal domain of the Pol II largest subunit (CTD), allowing cotranscriptional capping of the nascent pre-mRNA. Recent studies have indicated that CE affects transcription elongation and have suggested a checkpoint model in which cotranscriptional capping is a necessary step for the early phase of transcription. To investigate further the role of the CTD in linking transcription and processing, we generated a fusion protein of the mouse CTD with T7 RNA polymerase (CTD-T7 RNAP). Unexpectedly, in vitro transcription assays with CTD-T7 RNAP showed that CE promotes formation of DNA.RNA hybrids or R loops. Significantly, phosphorylation of the CTD was required for CE-dependent R-loop formation (RLF), consistent with a critical role for the CTD in CE recruitment to the transcription complex. The guanylyltransferase domain was necessary and sufficient for RLF, but catalytic activity was not required. In vitro assays with appropriate synthetic substrates indicate that CE can promote RLF independent of transcription. ASF/SF2, a splicing factor known to prevent RLF, and GTP, which affects CE conformation, antagonized CE-dependent RLF. Our findings suggest that CE can play a direct role in transcription by modulating displacement of nascent RNA during transcription.

Fig. 1.

T7 RNAP-CTD fusion proteins. T7 RNAP fused with mouse CTD. (A) Schematic representation of T7 RNAP fused with the CTD (CTD-T7 RNAP) and phosphorylated CTD (CTDp-T7 RNAP) are shown. (B) In vitro transcription by T7 RNAP and CTD-T7 RNAP (≈100 ng, respectively) with linearized plasmid DNA. Purified RNAs were resolved by denaturing PAGE. Runoff transcripts are indicated by arrow. (C) The 120-ng protein samples were resolved by SDS/PAGE and silver-stained. Molecular marker is indicated on the left. (D) In vitro transcription by CTDun-T7 RNAP and CTDp-T7 RNAP (≈120 ng, respectively) with linearized plasmid DNA.

Fig. 2.

Human CE promotes cotranscriptional RLF. (A) Template DNA for in vitro transcription driven by T7 promoter. Plasmid DNA was linearized by restriction enzyme digestion. The length of runoff transcripts is 295 nt. (B) The 1.2 μg of purified human CE was resolved by SDS/PAGE and visualized by Coomassie staining. (C) CE promotes the formation of HMW transcripts. In vitro transcription with CTDp-T7 RNAP was performed with increasing amounts (0, 60, 150, and 300 ng) of CE (lanes 1–4). Transcription by CTDun-T7 RNAP with increasing amounts (0, 60, 150, and 300 ng) of CE is shown in lanes 5–8. Transcription by T7 RNAP with increasing amounts (0, 60, and 300 ng) of CE is shown in lanes 9–11. CE-independent and CE-dependent HMW are indicated. The arrow indicates runoff transcripts. (D) HMW transcripts are DNA·RNA hybrids. Transcription of CTDp-T7 RNAP was performed in the presence (lanes 3 and 4) or absence (lanes 1 and 2) of 300 ng of CE. Transcription was terminated by heat (70°C, 15 min). Subsequently reaction mixtures were treated with (lanes 2 and 4) or without (lanes 1 and 3) RNase H, and RNAs were resolved by denaturing PAGE.

Fig. 3.

The GTase domain of human CE promotes RLF. (A) Experimental strategy for the RLF assay. Template strand (T), nontemplate strand (NT), and RNA are indicated. NT strand was labeled at its 5′ end (P). (B) RLF assay. Partially hybridized T and NT strands are shown (lane 2). Addition of RNA resulted in spontaneous RLF, converting ≈60% of NT into single-stranded NT (ssNT, lane 3). Heat-denatured products as molecular makers are shown in lane 1. Partially hybridized T and NT strands and ssNT are indicated on the right side. Asterisk indicates products made by aberrant T and NT hybridization. Top of the gel is indicated. (C) CE promotes RLF. RLF assays were performed with increasing amounts (0, 20, 50, 100, 200, and 400 ng) of CE (lanes 2–7) or 400 ng of BSA (lane 8). (D) Purified wild-type human GTase and GTase^(K294A) (2 μg, respectively) were resolved by SDS/PAGE and visualized by Coomassie staining. (E) RLF assays were performed with 200 ng of wild-type human GTase (hGTase, lane 3), hGTase^K294A (lane 4), or BSA (lane 5). (F) RLF assays were performed with 200 ng of each GTase (lane 2) and GTase-active mutant having a C-terminal truncation of 28 aa (lane 3). (G) In vitro transcription with CTDp-T7 RNAP was performed with increasing amounts (30, 60, 90, 120, and 150 ng) of hGTase (lanes 2–6) or hGTase^K294A (lanes 7–11). Samples were resolved by nondenaturing PAGE in B, C, E, and F and by denaturing PAGE in G.

Fig. 4.

Characterization of CE-dependent RLF. (A) ASF/SF2 prevents CE-dependent RLF. RLF assays were performed with 200 ng of hGTase in the presence of increasing amounts (0, 70, and 140 ng) of ASF/SF2 (lanes 3–5). The effect of 140 ng of ASF-SF2 alone is shown in lane 6. Putative triplet structures containing all three nucleic acid strands correspond to the lowest mobility products. (B) GTP inhibits CE-dependent RLF. RLF assays were performed with 200 ng of wild-type GTase alone (lane 4) or plus 1 mM GTP (lane 5); 1 mM GTP alone is shown in lane 3. (C) Wild-type, but not mutant, GTase is a target for GTP-mediated inhibition of RLF. RLF assays were performed with 200 ng of hGTase (lanes 2–10) or hGTase^K294A (lanes 11–19) in the presence of increasing amounts (0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.8, and 1.0 mM) of GTP. (D) Quantitation of DNA·RNA hybrids shown in C. The amounts of DNA·RNA hybrids caused by CE were quantitated by the ratio of hybridized template strands in the presence or absence of hGTase. (E) (Upper) RLF assays with 200 ng of hGTase plus 1 mM GTP, β,γ-imido-GTP, GDP, GMP, dGTP, and dATP were examined. (Lower) Quatitations of RLF are shown.

Fig. 5.

Possible roles of CE-dependent RLF during transcription in mammals. (A) The Spt5-containing DSIF complex, with NELF, functions to modulate transcriptional pausing at promoter-proximal regions. The DSIF complex and/or cap formation at the pre-mRNA 5′ end modulates RLF by CE. (B) Soon after promoter clearance, NELF dissociates from the Pol II EC, and CE acquires an ability to facilitate transient RLF to modulate elongation. (C) After transcription past the poly(A) site, RLF by CE induces transcriptional pausing. This finding could help to facilitate both cotranscriptional polyadenylation as well as subsequent transcription termination