A direct interaction between CPF and RNA Pol II links RNA 3' end processing to transcription

Abstract

Transcription termination by RNA polymerase II (RNA Pol II) is linked to RNA 3' end processing by the cleavage and polyadenylation factor (CPF or CPSF). CPF contains endonuclease, poly(A) polymerase, and protein phosphatase activities, which cleave and polyadenylate pre-mRNAs and dephosphorylate RNA Pol II to control transcription. Exactly how the RNA 3' end processing machinery is coupled to transcription remains unclear. Here, we combine in vitro reconstitution, structural studies, and genome-wide analyses to show that yeast CPF physically and functionally interacts with RNA Pol II. Surprisingly, CPF-mediated dephosphorylation promotes the formation of an RNA Pol II stalk-to-stalk homodimer in vitro. This dimer is compatible with transcription but not with the binding of transcription elongation factors. Disruption of the dimerization interface in cells causes transcription defects, including altered RNA Pol II abundance on protein-coding genes, tRNA genes, and intergenic regions. We hypothesize that RNA Pol II dimerization may provide a mechanistic basis for the allosteric model of transcription termination.

Keywords: CPSF; X-ray crystallography; cryo-EM; phosphatase; poly(A) tail; polymerase; transcription.

Graphical abstract

Figure 1

CPF and APT interact directly with RNA Pol II (A) Pull-down assay of CPF or APT using immobilized StrepII (SII)-tagged RNA Pol II (Rpb3-His-SII). Input and bound proteins were analyzed on SDS-PAGE. Labels show APT subunits (purple), core CPF subunits (black), and RNA Pol II (blue). The experiment was repeated twice. (B) (Top) Analytical size exclusion chromatography profiles of RNA Pol II (blue) and APT (purple), loaded separately or after incubation (black). APT with RNA Pol II-ΔCTD is in gray. (Bottom) SDS-PAGE (cropped) of the fractions indicated by a red line. The black dashed box indicates the migration position of the RNA Pol II-APT complex. Gels are outlined according to colors of chromatograms. (C) SDS-PAGE showing 3C protease cleavage of the Rpb1 CTD to make RNA Pol II-ΔCTD. (D) Analytical size exclusion chromatography experiment, as in (B), of RNA Pol II incubated with the Ref2-Glc7-Swd2 subcomplex of APT. See also Figures S1 and S2.

Figure 2

Cryo-EM analysis reveals RNA Pol II homodimers in the presence of Ref2-Glc7-Swd2 (A) 3D reconstruction of RNA Pol II-APT (gray transparent surface) with a model of RNA Pol II (PDB: 5C4X; blue ribbon) rigid-body fit into the map. DNA is in yellow, and RNA is in red. Density not accounted for is indicated. Arrows point at the RNA exit channel. (B and C) Selected 2D class averages of (B) DNA-RNA-loaded RNA Pol II with Ref2-Glc7-Swd2 from 850,000 particles and (C) DNA-RNA-loaded RNA Pol II from 400,000 particles. Classes are ordered by the number of particles in each class, from the most populated (1) to the least populated (9). (D) Composite map of the RNA Pol II dimer after signal subtraction and focused 3D refinement of the individual monomers (global resolution 3.6 Å). Monomers are colored in blue and gray. The DNA-RNA scaffold is in yellow. Anisotropy is due to preferred particle orientation on the grid. See also Figures S3–S6.

Figure 3

RNA Pol II dimerizes via the stalk protein Rpb7 (A) Two copies of monomeric RNA Pol II rigid-body fit into the dimeric RNA Pol II cryo-EM density. Monomer 1, blue; monomer 2, cyan; DNA, yellow. The magenta and green boxes on the model indicate the close-up views of the RNA Pol II dimer interfaces in (B) and (C), respectively. (B) Hydrophobic interaction between the Rpb7 subunits from each monomer. (C) Interaction between Rpb1 from monomer 1 with Rpb4 from monomer 2. (D–F) SDS-PAGE and cryo-EM analysis of RNA Pol II dimerization. 2D class averages were generated from 400,000 particles for RNA Pol II-Δstalk (D), RNA Pol II-Δstalk with recombinant wild-type (WT) stalk (E), or RNA Pol II-Δstalk with recombinant mutant stalk (F). Selected 2D classes are ordered by the number of particles per class, and dimeric classes are highlighted in red. Asterisks denote mutant stalk proteins. (G) Dynamic light scattering analysis on wild-type (dark gray) or Δstalk RNA Pol II (cyan). The hydrodynamic radius (R_h) is plotted as a function of RNA Pol II concentration. Black bars represent mean ± SD on 40 acquisitions for each concentration. The brackets on the right represent the increase in R_h from the lowest to the highest concentration. See also Figure S7 and Video S1.

Figure 4

Dephosphorylation promotes RNA Pol II dimerization (A) Selected 2D class averages of RNA Pol II-ΔCTD from 400,000 particles, ordered by the number of particles in each class. A dimeric class is highlighted in red. (B) Dynamic light scattering analysis on wild-type (dark gray; replotted from Figure 3G) or RNA Pol II-ΔCTD (green). Black bars represent mean ± SD on 40 acquisitions for each concentration. The brackets on the right represent the increase in R_h between the lowest and highest concentrations for each condition. (C) Analysis of RNA Pol II dimerization by negative stain EM. The percentage of dimeric particles per micrograph is plotted. For each RNA Pol II treatment, 60 (left) or 50 (middle and right) micrographs were acquired. The experiments in the left and middle panels were performed twice, once as a “double-blind” experiment. Black bars represent the mean ± SD. Pairwise comparisons are shown as indicated, where ^∗∗∗∗p < 0.0001 by one-way ANOVA Tukey’s test. (D) Schematic of regions used to calculate the RNA Pol II retention index. TSS, transcription start site; TES, transcription end site; pc, protein-coding. (E) Scatter plot of the log₂ fold change in RNA Pol II retention index versus the log₂-fold change in RNA Pol II signal across the transcribed unit for genes with annotated transcription start and end sites (5,357 genes). Data points highlighted in red correspond to genes with RNA Pol II retention at the 3′ end (clusters 2–4 from the subset of data analyzed in F). The percentage of genes within each retention index bracket is shown. (F) k-means clustering and heatmap of RNA Pol II occupancy beyond the transcription end site (TES) at mRNA genes with RNA Pol II signal ≥0.5 in both wild-type (WT) and Rpb7^QHF cells. Clusters 2–4 show RNA Pol II retention in Rpb7^QHF cells compared with WT. See also Figures S8 and S9.

Figure 5

The Ref2 subunit of CPF and APT is a regulatory subunit of Glc7 (A) Schematic of the Glc7-Ref2_348–406 chimeric protein used for crystallization. Orange, Glc7; blue, Ref2_348–406; gray, glycine-serine linker; dark blue, two SPTYSPS RNA Pol II CTD repeats. Black lines indicate the regions visible in the crystal structure. (B) Cartoon representation of the Glc7-Ref2_348–406 crystal structure in two orientations and close-up views of the interactions of Ref2 residues 358–362 (left) and of the intermolecular β sheet formed between Glc7 and Ref2, including interaction details for the hydrophobic pair (ΦΦ) (right). Two manganese ions and a phosphate group are shown in ball-and-stick. Coordination waters are in red. The N and C termini are indicated. (C) View of the interaction interface between Ref2_348–406 in cartoon and Glc7 in surface representation. The inset shows how Ref2 binds Glc7 through the conserved “RVxF” motif. (D) Sequence alignment of yeast Ref2 and putative orthologs from Rattus norvegicus (Rn) and Homo sapiens (Hs). The Ref2 “I/L-x-R-x-G-K/R” motif is enclosed in a light blue box. The RVxF and downstream ΦΦ motifs shared among PP1-regulators are highlighted in light blue. (E) Immunoblots (top, anti-FLAG; bottom, anti-α-tubulin) of Ref2-mAID cells before or after addition of 1 mM auxin (IAA). Ref2-mAID is degraded within 15 min of adding auxin. The anti-FLAG antibody cross-reacts with a protein (^∗) also present in the wild-type parent strain (final lane). n = 3. (F) Growth curves of Ref2-mAID (left) or Ref2-mAID cells co-expressing a triple-point mutant of Ref2 (Ref2^mut: I372D, F374K, Y384E). Cells were grown with 1 mM auxin (IAA) or an equivalent volume of DMSO. Dotted line, mean OD₆₀₀ of biological replicates (n = 3); shaded area, standard deviation of the mean. See also Figure S10.

Figure 6

Ref2 is required for the APT interaction with RNA Pol II (A) Pull-down assay of APT and APT-ΔRef2 with RNA Pol II immobilized on StrepTactin beads. SII, StrepII-tagged protein; APT subunits, purple; RNA Pol II subunits, blue; asterisk, degradation product. APT-ΔRef2 was obtained after Ref2 was degraded by contaminating proteases during purification of APT. (B) Analytical size exclusion chromatography of RNA Pol II and Ref2-ΔIDR-Glc7-Swd2. The fractions indicated by a red line were analyzed on the SDS-PAGE below. See also Figure S11.

Figure 7

Proposed model for the role of CPF in transcription termination (A) CPF and APT can bind elongating RNA Pol II to monitor RNA as it emerges from the RNA exit channel. PAS sequence, orange box; cleavage site, gray box; nascent RNA, dark red; yellow circles, phosphorylation. Transcription elongation factors Spt4/5 are also shown. (B) After the PAS is transcribed, CPF-mediated pre-mRNA cleavage and RNA Pol II CTD-dephosphorylation are activated. As a result, the newly exposed 5′ end still attached to RNA Pol II is degraded by the torpedo exonuclease Rat1 (left), and RNA Pol II dephosphorylation triggers an allosteric event (RNA Pol II dimerization), which displaces transcription factors, or domains of transcription factors (right). Thus, CPF may convert RNA Pol II into a termination-competent complex. It remains unclear whether the second RNA Pol II is also transcribing. See also Figure S12.