Recruitment, Duplex Unwinding and Protein-Mediated Inhibition of the Dead-Box RNA Helicase Dbp2 at Actively Transcribed Chromatin

Abstract

RNA helicases play fundamental roles in modulating RNA structures and facilitating RNA-protein (RNP) complex assembly in vivo. Previously, our laboratory demonstrated that the DEAD-box RNA helicase Dbp2 in Saccharomyces cerevisiae is required to promote efficient assembly of the co-transcriptionally associated mRNA-binding proteins Yra1, Nab2, and Mex67 onto poly(A)(+)RNA. We also found that Yra1 associates directly with Dbp2 and functions as an inhibitor of Dbp2-dependent duplex unwinding, suggestive of a cycle of unwinding and inhibition by Dbp2. To test this, we undertook a series of experiments to shed light on the order of events for Dbp2 in co-transcriptional mRNP assembly. We now show that Dbp2 is recruited to chromatin via RNA and forms a large, RNA-dependent complex with Yra1 and Mex67. Moreover, single-molecule fluorescence resonance energy transfer and bulk biochemical assays show that Yra1 inhibits unwinding in a concentration-dependent manner by preventing the association of Dbp2 with single-stranded RNA. This inhibition prevents over-accumulation of Dbp2 on mRNA and stabilization of a subset of RNA polymerase II transcripts. We propose a model whereby Yra1 terminates a cycle of mRNP assembly by Dbp2.

Keywords: DEAD-box; RNA; RNA–protein complex; chromatin; helicase.

Figure 1. Dbp2 is recruited to chromatin via RNA

(A) Schematic diagram of the GAL10 and GAL7 genes and the positions of qPCR amplicons. (B) Dbp2 is recruited to chromatin in an RNA-dependent manner. Transcription of the GAL genes was induced by growing yeast cells in rich media plus glucose initially and subsequently shifting to media with galactose for 5 hours. Chromatin was then isolated, sheared by sonication, and incubated with 7.5 U RNase A and 300 U RNase I or buffer alone before being subjected to ChIP using anti-FLAG antibodies. Results are shown as the percent of precipitated DNA over input averaged across four biological replicates with SEM. * indicates a p-value < 0.05 as assessed by student's t-test. (C) Diagram of the functional motifs of Yra1 [22,25,45,80]. The C-terminal half of Yra1 interacts with Dbp2 in vitro [17]. (D) Dbp2 interacts with the C-terminal half of Yra1 in vivo. Immunoprecipitation assays were conducted using anti-FLAG antibodies to isolate Dbp2-3xFLAG and associated proteins from wild type or yra1ΔC strains. 10% of the lysate was used as input. Dbp2 and Yra1 were detected by Western blotting with protein-specific antibodies. Immunoprecipitated Dbp2, Yra1 and yra1ΔC were quantified by densitometry and are shown as a percentage of input. (E) Loss of the C-terminal half of Yra1 does not affect the association of Dbp2 with GAL10 (left) or GAL7 (right) genes. WT and yra1ΔC strains were used for ChIP with anti-FLAG antibody against Dbp2-3xFLAG. Student t-test was performed between full-length YRA1 and yra1ΔC strains in all primer sets. All the p-values > 0.05.

Figure 2. Yra1 prevents over-accumulation of Dbp2 on RNA Pol II transcripts

(A) Dbp2 accumulates on the RNA Pol II transcripts in a yra1ΔC strain. RNA immunoprecipitation (RIP) assays were performed to determine the level of RNA associated with Dbp2 in wild type and isogenic yra1ΔC cells. Cells were grown with galactose to promote expression of GAL10 and GAL7 genes as in Fig. 1 and subsequently cross-linked with formaldehyde. RNPs were isolated with anti-FLAG antibodies and transcripts were detected by RT-qPCR with primers specific to the 5’ end of each mRNA (see Supplemental Table 4). Dbp2-3xFLAG occupancy on specific transcripts is shown as the average percent of isolated RNA over input for three biological replicates. Error bars indicate the SEM. (B) The association of Dbp2 with RNA Pol II transcripts is not altered in the mRNA export mutant strain, rat7-1. RIP assays were performed as above with wild type cells, isogenic rat7-1 cells [81], or isogenic, wild type untagged cells at both the permissive temperature (25°C, left) and the non-permissive temperature (37°C, right) for rat7-1 [–28].

Figure 3. Dbp2 forms a large RNA-dependent complex with Yra1 and Mex67 in vivo

(A) Mex67 and Yra1 copurify with immunoprecipitated Dbp2. TAP-tag immunoprecipitation assays of Dbp2 were conducted in the presence or absence of RNase A and RNase I. Input (1%) and elutions were resolved by SDS-PAGE and proteins were detected by western blotting. Cross-reactive heavy (HC) and light chains (LC) are noted whereas an asterisk (*) marks a nonspecific band. Note that the faster migrating band in lanes 4-5 corresponds to LC not untagged Dbp2, as there is no untagged DBP2 expressed in the haploid, DBP2-TAP strain. (B) Dbp2, Yra1 and Mex67 co-migrate as a large complex by glycerol gradient fractionation. Glycerol gradient (10 – 30%) were performed with yeast lysate and the isolated fractions were resolved by SDS-PAGE and proteins were detected by western blotting. Molecular weights were determined using a standard curve generated by resolving catalase (250 kDa), apoferritin (480 kDa), and thyroglobulin (670 kDa). (C) RNase treatment of yeast lysate prior to gradient fractionation shifts the migration pattern of Dbp2, Yra1 and Mex67. Glycerol gradient (10 – 30%) were performed as above but with RNase A and RNase I.

Figure 4. Yra1 decreases the number of RNA hairpins unwound by Dbp2 without altering the kinetics of unwinding

(A) Schematic representation of smFRET with a doubly-labeled hairpin RNA. Dual labeled RNA (Cy3 and Cy5) was purchased from IDT and subsequently surface-immobilized on a pegylated microscope quartz slide via biotin-neutravidin bridge (shown as yellow and orange ovals) [82]. The oval represents Dbp2. The red star represents Cy5 and the green star represents Cy3. (B) Representative FRET trajectory in the smFRET experiments. Representative trajectories of a closed RNA hairpin alone (top left), in the presence of 10 nM Dbp2 (top right),10 nM Dbp2 and 100 μM ATP (bottom left), or in the presence of 10 nM Dbp2, 20 nM yra1C, and 100 μM ATP (bottom right) are shown. (C) Yra1 decreases the number of hairpin dsRNAs unwound by Dbp2. The distribution of closed, closed to opened (single opening events), or dynamic (multiple cycles of opening and closing) hairpin dsRNAs with 10 nM Dbp2 with or without 100 μM hexokinase and 1 mM glucose in the absence of ATP or 10 nM Dbp2 with increasing concentrations of GST-yra1C in the presence of ATP are shown. Trajectories exhibiting more than one excursion into 0.2 – 0.8 FRET, i.e., opening more than once, are considered dynamic molecules. Trajectories exhibiting constant high (0.9) or low (0.1) FRET throughout the experimental time window are classified as closed or opened molecules, respectively. A threshold of 0.6 FRET was used to distinguish between the open (0.1 FRET) and close (0.9 FRET) state of the hairpin. (D) Yra1 does not appreciably alter the kinetics of Dbp2-dependent unwinding. Dwell times for each opening/closing event were measured from FRET time trajectories from dynamic dsRNA molecules. Single exponential decay was used to fit the dwell time histograms to determine k_opening and k_closing respectively.

Figure 5. Yra1 reduces the RNA-binding affinity of Dbp2 in vitro

(A) Yra1C reduces the affinity of Dbp2 for RNA. Fluorescence anisotropy assays were conducted with varying amounts of Dbp2 with or without 150 nM of GST-yra1C and 10 nM of a 16mer fluorescently labeled ssRNA (5’-6-FAM-AGC ACC GUA AAG ACG C-3’) in the presence or absence of 2 mM ADP-BeF_x under equilibrium conditions. Experiments were conducted in triplicate with the average shown and error bars indicating the SEM. (B) Yra1C decreases association of Dbp2-ADP-BeFx with RNA in pre-equilibrium experiments. Gel shift assays were conducted in the presence of 2 mM ADP-BeF_X/MgCl₂, 10 nM of a 5’-radioactively labeled ssRNA (5’-AGC ACC GUA AAG ACG C-3’), with or without the Dbp2 (400 nM) and varying amounts of GST-yra1C or BSA (0 nM, 300 nM, 600 nM, 1200 nM, and 1800 nM). Complexes were assembled at 4°C as indicated in the schematic diagram followed by resolution on a 4% native PAGE and subsequent autoradiography. Time courses show that it takes >60 min to reach equilibrium (Supplemental Fig. 3). ND indicates the protein-bound signal was not detected.

Figure 6. The Dbp2-Yra1 interaction prevents overexpression of specific of gene products in vivo

(A) Loss of the Dbp2-Yra1 interaction results in overaccumulation of GAL7 and DBP2 transcripts. Growing wild type or yra1ΔC cells with galactose induced transcription of the GAL genes. RT-qPCR was performed for the indicated genes following extraction of RNA using primers listed in Supplemental Table 4. Transcript levels were normalized to 18S rRNA and wild type. Error bars indicate the SEM from three biological replicates and * indicates a p-value <0.05 from a two tailed student t-test. (B) Loss of the Dbp2-Yra1 interaction increases protein levels of Gal7 and Dbp2. C-terminally 3X-FLAG-tagged GAL10 and GAL7 strains were constructed in the yra1ΔC strain by standard yeast methods to provide an epitope for western blotting. Protein detection by western blotting was conducted using anti-Dbp2 [23] or anti-FLAG as indicated. Protein signal intensity was quantified with ImageQuant. The average signal and SEM with respect to Pgk1 is shown for three biological replicates. (C) RNA Pol II exhibits a similar pattern of gene occupancy in both wild type and yra1ΔC strains. ChIP was performed as above, but with anti-Rpb3 antibodies, a subunit of RNA Pol II. (D) The GAL7 mRNA has a longer half-life in yra1ΔC strains than wild type cells. Transcriptional shut off assays were performed by shifting indicated strains to glucose to repress transcription of GAL genes after a 10 hours induction with galactose. RNA was extracted at the indicated time points and transcripts were detected by Northern blotting. Transcripts were quantified by densitometry and normalized to scR1. Half-lives were calculated from three, independent biological replicates by fitting the data to an exponential decay equation.

Figure 7. Enzymatic inhibition of Dbp2 by Yra1 restricts cycles of Dbp2-dependent mRNP remodeling in vivo

Dbp2 is co-transcriptionally recruited to chromatin through RNA to resolve RNA duplexes. This resolution allows co-transcriptional loading of RNA-binding proteins Yra1, Mex67, and Nab2 onto the nascent RNA. After nucleotide exchange, Yra1 prevents post-transcriptional re-association by reducing the single-stranded RNA binding affinity of Dbp2. This activity likely prevents Dbp2 from accumulating on mRNA, which results in aberrant transcript stabilization and overexpression of specific gene products.