The Nrd1-Nab3-Sen1 termination complex interacts with the Ser5-phosphorylated RNA polymerase II C-terminal domain

Abstract

RNA polymerase II (Pol II) in Saccharomyces cerevisiae can terminate transcription via several pathways. To study how a mechanism is chosen, we analyzed recruitment of Nrd1, which cooperates with Nab3 and Sen1 to terminate small nucleolar RNAs and other short RNAs. Budding yeast contains three C-terminal domain (CTD) interaction domain (CID) proteins, which bind the CTD of the Pol II largest subunit. Rtt103 and Pcf11 act in mRNA termination, and both preferentially interact with CTD phosphorylated at Ser2. The crystal structure of the Nrd1 CID shows a fold similar to that of Pcf11, but Nrd1 preferentially binds to CTD phosphorylated at Ser5, the form found proximal to promoters. This indicates why Nrd1 cross-links near 5' ends of genes and why the Nrd1-Nab3-Sen1 termination pathway acts specifically at short Pol II-transcribed genes. Nrd1 recruitment to genes involves a combination of interactions with CTD and Nab3.

Figure 1

Structure of the Nrd1 CID. (a) The CID regions of Nrd1 proteins from several yeast species were aligned with ClustalW. Identical amino acid residues are shown in dark green, and declining sequence similarity is shown using light green, orange and yellow, in that order. Helices within the Nrd1 CID model are depicted in blue above the sequence alignment. Loop regions are shown in beige, and the loop region missing in the Nrd1 CID model in cyan. (b) Ribbon illustration of the Nrd1 CID polypeptide chain. The bound sulfate ion contacting the helix 1-helix 2 loop is represented as a stick model, and the missing loop between helices 4 and 5 is modeled as random coil in cyan. (c) Surface representation of Nrd1 CID colored according to the sequence conservation as in a. The bound sulfate ion is represented as a stick model. (d) Ribbon model of the Pcf11 CID domain (PDB 1SZA), with bound CTD-Ser2P represented as a stick model.

Figure 2

Nrd1 binds preferentially to CTD-Ser5P. (a) Binding to four repeat CTD peptides in vitro. Unmodified, Ser2P, Ser5P or Ser2P/Ser5P peptides were immobilized on streptavidin-conjugated magnetic beads and incubated with 5 µg of recombinant Nrd1 (rNrd1), 10 ng of TAP-purified yeast Nrd1 complex (yNRD1) or 500 µg of whole-cell extract from an Rtt103-hemagglutinin (HA) strain (YSB815). Bound proteins were eluted, separated by SDS-PAGE and detected by immunoblotting using either anti-Nrd1 or anti-HA antibodies. Recombinant Rtt103 also specifically bound to CTD-Ser2P (not shown). (b) Nrd1_6–151 was titrated with fluorescently labeled CTD-Ser5P (two repeats) and binding was measured by fluorescence anisotropy (black triangles; ref.-FAM, 5,6-carboxyfluorescein labeled reference). The same experiment was then done in the presence of competing unlabeled CTD-Ser2P (circles), CTDSer5P (white triangles) or CTD-Ser2P/Ser5P (diamonds). Equilibrium dissociation constants (K_d) were calculated from the best fit to the data. (c) Nrd1 is associated with Ser5-phosphorylated Pol II in vivo. Nrd1 was purified via the TAP tag, and the phosphorylation status of the associated polymerase was monitored by immunoblotting using anti-CTD (8WG16), anti-Ser2P (H5), anti-Ser5P (H14) or an antibody that can recognize both Ser2P and Ser5P (B3). (d) Ctk1 kinase is not required for recruitment of Nrd1 to genes in vivo. Cross-linked chromatin was prepared from Nrd1-TAP–containing cells that were wild-type (WT) or deleted (Δctk1) for the CTK1 gene. Following precipitation with IgG agarose, chromatin was amplified with primers across the snR33 locus, as diagrammed below. Immunoprecipitated samples (IP) were compared against input chromatin (Input) and quantified (right). The upper band in each lane is the snR33 product and the lower band is a nontranscribed control region. Similar results were obtained for the PMA1 and ADH1 genes (not shown).

Figure 3

Structural models for a Nrd1-CTD interaction. Left, the CTD-Ser2P peptide is modeled bound to Nrd1 by simple manual superposition of the Nrd1 CID onto the Pcf11 CTD. Right, the longer CTD model was created by superimposing the extended CTD-Ser5P bound to mRNA capping enzyme (PDB 1P16) onto the extended region of the CTD-Ser2P bound to Pcf11. The overlapping stretch of the two CTDs consists of Ser7-Tyr1-Ser2. Notably, the position of the phosphate moiety from Ser5P coincides with the bound sulfate ion observed in the Nrd1 crystal structure. Blue arrows point out phosphorylated CTD residues; black arrows show residues mutated and tested for effects on CTD binding in Table 1.

Figure 4

The CTD and Nab3 interaction domains of Nrd1 are both important for interaction with Pol II. (a) Schematic diagram of Nrd1. RE/RS, arginine-, serine- and glutamate-rich region; P/Q, proline- and glutamine-rich region. (b) Phenotypic analysis of nrd1Δ6–214 and nrd1Δ151–214 deletions. The NRD1 plasmid shuffling strain EJS101-9d was transformed with pJC580, pRS415-Nrd1Δ6–214 or pRS415-Nrd1Δ151–214, and the wild-type NRD1/URA3 plasmid (pRS316-NRD1) was shuffled out on 5-fluoroorotic acid (5-FOA) medium at the indicated temperatures. (c) Phenotypic analysis of nrd1Δ6–150 and nrd1Δ151–214 alleles integrated into the genome. (d) Nrd1 residues 151–214 are responsible for interaction with Nab3. Expression of the wild-type (WT) and mutant TAP-tagged proteins Nrd1, Nrd1Δ6–214 and Nrd1Δ151–214 in extracts was monitored by immunoblotting for the Protein A module of the TAP tag (below, α-TAP). Nrd1 protein complexes were purified using IgG resin, and association with Nab3 was analyzed with anti-Nab3 antibody (above). Note that the Protein A module on Nrd1 reacts with the secondary antibody. (e) The Nab3 and CTD binding regions of Nrd1 contribute to its interaction with Pol II in vivo. Nrd1, Nrd1Δ6–150, Nrd1Δ151–214 or Nrd1Δ6–214 protein complexes were IgG-purified from whole-cell extracts and monitored for association with Pol II by immunoblotting with anti-CTD Ser5P (H14).

Figure 5

Efficient Nrd1 recruitment to 5′ ends of Pol II–transcribed genes requires both the CID and Nab3 interaction domain. (a) Deletion of Nrd1 residues 151–214 reduces Nrd1 recruitment to the SNR13 gene. Schematic representation of snR13 is shown above. Coding regions are shown as boxes; arrows indicate promoters; numbered bars show positions of the PCR products used in ChIP analysis; asterisks indicate a control band amplified from a nontranscribed region of Chromosome V. ChIP (IP) results are shown in the upper panels, and PCR from non-IPed chromatin samples is shown below (Input). Quantification of the ChIP data are shown to the right. The same chromatin preparations were used to analyze Pol II subunit Rpb3. The y-axis of the graph shows Nrd1 levels as a ratio to Rpb3 levels. The x-axis refers to the numbered primer pairs. (b,c) ChIP analysis of full-length Nrd1, Nrd1 lacking the CID (Δ6–150), or Nrd1 lacking the Nab3 interaction region (Δ151–214) was carried out on the PMA1 and PYK1 genes as in a. Graphs show the levels of Nrd1 normalized to Rpb3 levels.

Figure 6

Effect of Nrd1 deletions on snoRNA termination and processing. (a) Schematic diagram of snR13 and snR33. Coding regions are shown as boxes, transcription start sites as bent arrows, and transcription termination regions are designated as STOP. The positions of the probes are shown above the gene. (b) Northern blot analysis of snR13 and snR33 RNA. Total RNA was isolated from NRD1, nrd1Δ6–150 and nrd1Δ151–214 cells and analyzed as described previously,. Positions of the transcription read-through transcripts, unprocessed precursor and mature snoRNAs are indicated with arrows; a previously observed truncated species is marked with an asterisk. (c) Northern blot analysis of snR13 RNA from cells expressing Nrd1, Nrd1Δ151–214 and Nrd1Δ6– 214 proteins. In this experiment, wild-type Nrd1 is expressed from a galactose-inducible promoter, and depletion occurs upon shift to glucose media. The Nrd1 deletions are expressed from plasmids. The ‘−’ panel shows cells carrying only vector. (d) Expression of the snR33 gene was analyzed as in c.