Author response: Structural mechanism of ATP-independent transcription initiation by RNA polymerase I

Article Figures and data Abstract Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Transcription initiation by RNA Polymerase I (Pol I) depends on the Core Factor (CF) complex to recognize the upstream promoter and assemble into a Pre-Initiation Complex (PIC). Here, we solve a structure of Saccharomyces cerevisiae Pol I-CF-DNA to 3.8 Å resolution using single-particle cryo-electron microscopy. The structure reveals a bipartite architecture of Core Factor and its recognition of the promoter from −27 to −16. Core Factor's intrinsic mobility correlates well with different conformational states of the Pol I cleft, in addition to the stabilization of either Rrn7 N-terminal domain near Pol I wall or the tandem winged helix domain of A49 at a partially overlapping location. Comparison of the three states in this study with the Pol II system suggests that a ratchet motion of the Core Factor-DNA sub-complex at upstream facilitates promoter melting in an ATP-independent manner, distinct from a DNA translocase actively threading the downstream DNA in the Pol II PIC. https://doi.org/10.7554/eLife.27414.001 Introduction Eukaryotic RNA synthesis is catalyzed by at least three classes of RNA Polymerases (Pol I-III) (Roeder and Rutter, 1969). The large ribosomal RNA precursor (pre-rRNA) is transcribed by Pol I (Moss et al., 2007), accounting for up to 60% of total cellular RNA synthesis in Saccharomyces cerevisiae (Warner, 1999). Transcription by Pol I is highly regulated, and its mis-regulation has been implicated in many diseases including various types of cancer (Drygin et al., 2010; Montanaro et al., 2013; White, 2008). Pre-Initiation Complex (PIC) formation is a key regulatory step in the control of gene transcription by eukaryotic RNA polymerases. Yeast Pol I transcription initiation is regulated by four general transcription factors: the regulatory factor Rrn3, the Core Factor (CF), the TATA-box Binding Protein (TBP), and the Upstream Activation Factor (UAF) (Schneider, 2012). Rrn3 contains an elongated HEAT repeat (Blattner et al., 2011), and binds Pol I via contacts with subunits A43, A190, and AC40 (Blattner et al., 2011; Cavanaugh et al., 2008; Engel et al., 2016; Milkereit and Tschochner, 1998; Peyroche et al., 2000; Pilsl et al., 2016). Rrn3 association stabilizes Pol I in its monomeric and initiation-competent form (Blattner et al., 2011; Engel et al., 2016; Pilsl et al., 2016; Torreira et al., 2017), with which Core Factor further engages to facilitate Pre-Initiation Complex assembly and transcription initiation (Aprikian et al., 2001; Knutson and Hahn, 2013; Milkereit and Tschochner, 1998; Peyroche et al., 2000; Schneider, 2012). In addition to recruiting Pol I/Rrn3 to the ribosomal DNA (rDNA) promoter, Core Factor has also been implicated in transcription bubble opening (Kahl et al., 2000). Core Factor recruitment to the rDNA promoter in vivo requires the association of UAF with the upstream activating sequence (UAS) and TBP (Bordi et al., 2001; Oakes et al., 1999; Steffan et al., 1996; Vannini, 2013). Transcription initiation by the three eukaryotic RNA RNA polymerases requires transcription factor (TF) IIB-like factors (Vannini, 2013; Vannini and Cramer, 2012). TFIIB binds the Pol II dock and wall domains using its N-terminal zinc ribbon (ZR) (Bushnell et al., 2004; Chen and Hahn, 2004) and C-terminal cyclin fold domains (Chen and Hahn, 2004; Kostrewa et al., 2009; Sainsbury et al., 2013), respectively. The Rrn7 subunit of Core Factor is predicted to share sequence homology with TFIIB (Blattner et al., 2011; Knutson and Hahn, 2011; Naidu et al., 2011), containing similar ZR and cyclin fold domains in addition to a helical C-terminal domain (CTD). Models for the Pol I Pre-Initiation Complex were proposed based on the similarity between TFIIB and Rrn7 (Blattner et al., 2011; Knutson et al., 2014), which have recently been challenged by the crystal structure of Core Factor (Engel et al., 2017). In addition to Rrn7, Rrn6 and Rrn11 are essential subunits of Core Factor (Lalo et al., 1996; Lin et al., 1996). The human ortholog of Core Factor is Selectivity Factor 1 (SL1), which comprises three evolutionarily conserved core subunits and two additional metazoan-specific subunits, TAF1D and TAF12 (Denissov et al., 2007; Gorski et al., 2007), suggesting a conserved architecture between these Pol I general transcription factors (Knutson and Hahn, 2013; et al., 2006; Schneider, 2012). Structural approaches have elucidated the dynamic nature of Pol I, possibly reflecting potential conformational states that it can adopt during different stages of transcription. First, in the atomic structure of Pol I determined by X-ray crystallography, a dimeric configuration and an expanded DNA-binding cleft were observed (Engel et al., 2013; Fernández-Tornero et al., 2013). The DNA-binding cleft is occupied by an element named the expander, mimicking a DNA molecule, while another element called the connector contributes to the dimerization interface by engaging the clamp domain of the neighboring Pol I (Engel et al., 2013; Fernández-Tornero et al., 2013). In addition, the bridge helix at the active site partially unfolds. When interacting with Rrn3, both the expander and the connector are displaced, resulting in a monomeric form of Pol I, with a more contracted cleft and a partially rewound bridge helix (Engel et al., 2016; Pilsl et al., 2016). A further contraction of the cleft and a completely folded bridge helix were observed in the elongation form of Pol I revealed by cryo-EM (Neyer et al., 2016; Tafur et al., 2016). Although these studies provided intriguing hints at the mechanisms of Pol I transcription initiation and elongation, the lack of a Pre-Initiation Complex in these studies precluded a full understanding of its engagement with the promoter and its transition to an active transcribing state. To gain insight into Core Factor's role during Pol I transcription initiation, we obtained a Pol I Initial Transcribing Complex (ITC) and determined its structure to near-atomic resolution using cryo-EM. In particular, we describe three distinct functional states of the Pol I initiation complexes visualized at 3.8–4.3 Å resolution. Our structures reveal unexpected features of Core Factor's binding to Pol I and promoter DNA compared to the Pol II Pre-Initiation Complex, and provide novel insight into the mechanism of Pol I promoter opening utilizing the intrinsic mobility of Core Factor in the absence of ATP hydrolysis Results Assembly and cryo-EM reconstruction of the Pol I initiation complex on promoter DNA To gain insight into the regulation of Pol I transcription initiation, we assembled the Pol I basal transcription complex on an rDNA promoter using purified factors from Saccharomyces cerevisiae (Materials and methods; Figure 1—figure supplement 1). In order to stabilize the complex, we used a nucleic acid scaffold containing a 17-nucleotide (nt) mismatched transcription bubble in the presence of a 6-nt RNA molecule, mimicking an initial transcribing state (Figure 1A). Figure 1 with 4 supplements see all Download asset Open asset Cryo-EM structure of Pol I Initial Transcribing Complex. (A) Nucleic acid scaffold used. The non-template and template strands are depicted in cyan and blue, respectively. Filled circles represent rDNA promoter sequence, while open circles show the poly-T mismatch sequences. RNA is shown in red. Core Factor binding region is also labeled. (B) Cryo-EM reconstruction of Pol I Initial Transcribing Complex following focused refinements on Core Factor and Pol I separately (Materials and methods). Pol I is colored gray, and nucleic acid template is colored as shown in A. The Core Factor subunits are depicted in pink (Rrn6), green (Rrn7) and gold (Rrn11). Two views, front (left) and bottom (right), are shown. (C) MDFF (molecular dynamics flexible fitting) model of the Pol I Initial Transcribing Complex. Components are colored the same as in B. https://doi.org/10.7554/eLife.27414.002 Single particle analysis using RELION (Kimanius et al., 2016; Scheres, 2012) produced a reconstruction with an overall resolution of 3.8 Å (FSC = 0.143 criterion) (Figure 1B, Figure 1—figure supplement 2, Video 1). The reconstruction shows a bipartite configuration, with the large module showing clear features of Pol I and the small lobe corresponding to Core Factor. DNA density was observed both inside and upstream of the Pol I cleft (Figure 1B), with the upstream DNA interacting with Core Factor. Local resolution estimation shows that the Pol I core region is very rigid with resolution mostly at 3.5 Å, whereas peripheral regions such as the stalk and Core Factor are more mobile with a lower resolution (4.5 to 5.5 Å) (Figure 1—figure supplement 2C). This indicates a flexible nature of the Core Factor in the Initial Transcribing Complex. Indeed, maximum-likelihood based 3D classification revealed that Core Factor adopts different orientations relative to Pol I, likely suggesting a continuous motion for Core Factor. Unique features were also observed accompanying the movement of Core Factor, which will be discussed in a subsequent section. Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Cryo-EM reconstruction and MDFF model of Pol I Initial Transcribing Complex. Densities are shown as a semi-transparent surface following a similar color scheme to that in Figure 1. https://doi.org/10.7554/eLife.27414.007 To obtain a higher resolution density map, especially for the region corresponding to the Core Factor-DNA interaction, we applied separate soft masks around the Core Factor and Pol I densities and performed focused refinements for both individually (Figure 1—figure supplement 2C). This resulted in improved density maps for both Pol I and Core Factor, with an overall resolution of 3.7 Å and 4.2 Å, respectively (Figure 1—figure supplement 2C,D). This procedure permitted de novo model building for Core Factor (Figure 1C). We also generated and refined an atomic model for Pol I based on the crystal structures (Engel et al., 2013; Fernández-Tornero et al., 2013). Compared to similar structures determined by either crystallography in the apo form (Engel et al., 2013; Fernández-Tornero et al., 2013) or cryo-EM in an Elongation Complex (EC) (Neyer et al., 2016; Tafur et al., 2016), Pol I within our Initial Transcribing Complex reconstruction resembles the Elongation Complex more than the apo crystal structure, with a contracted active site cleft (Figure 1—figure supplement 3). In addition, Core Factor engages Pol I in the vicinity of the protrusion domain near the upstream entrance of the DNA binding cleft, and intimately interacts with the upstream promoter DNA (Figure 1B). Although we included both Rrn3 and TBP in our assembly reactions (Materials and methods), we did not observe densities in any of the classified reconstructions that could correspond to them (Figure 1—figure supplement 2). Therefore, we monitored all fractions of our assembly reaction by gel electrophoresis (Figure 1—figure supplement 4). Consistent with the absence of Rrn3 in our reconstructions, we found out that all of the Rrn3 protein was in the unbound and the first wash fractions (lanes 5 and 6 in Figure 1—figure supplement 4). Although we cannot rule out the possibility that Rrn3 is present at a substoichiometric level that is below the detection limit of silver staining, or alternatively, that Rrn3 associates with Pol I in the unbound fraction that somehow failed to engage the nucleic acid scaffold, our data suggests that Rrn3 does not stably associate with the rest of Pol I initiation machinery under our experimental conditions. Given the essential roles Rrn3 plays during Pol I transcription (Keener et al., 1998; Moorefield et al., 2000; Schneider, 2012; Yamamoto et al., 1996) and its dissociation from Pol I after transcription initiation (Bier et al., 2004; Hirschler-Laszkiewicz et al., 2003; Milkereit and Tschochner, 1998), our data is consistent with the notion that Rrn3 functions at an earlier step during Pol I transcription initiation, where it stabilizes Pol I in an initiation-competent monomeric form and facilitates Pol I recruitment to rDNA promoter. In a separate study, Reeder and colleagues reported that Pol I can be recruited to the promoter in the absence of Rrn3, however this complex is inactive (Aprikian et al., 2001), suggesting that Rrn3 may also function post Pol I recruitment. This is also consistent with our structural study, as we included an RNA molecule in our bubble template, which may have resulted in bypassing the requirement for Rrn3 after the engagement of Pol I with the nucleic acid scaffold. As for TBP, we observed a band in the elution fraction that could result from TBP non-specifically binding to the DNA template (Figure 1—figure supplement 4, compare lanes 9 and 12), consistent with the absence of TBP in our structures. Molecular structure of Core Factor The overall structure of DNA-bound Core Factor resembles a right hand holding the DNA molecule between the fingers and the palm, with the thumb pointing toward Pol I (Figure 2—figure supplement 1A). The palm is composed of the N-terminal regions of both Rrn11 and Rrn6, the thumb is composed of the C-terminus of Rrn11, and the fingers and knuckles are composed of Rrn7 and the C-terminal half of Rrn6, respectively (Figure 2—figure supplement 1A). Rrn6 plays a scaffolding role in the assembly of Core Factor, spanning the palm and the knuckles (Figure 2—figure supplement 1A). As predicted (Knutson et al., 2014), Rrn6 is composed of an N-terminal domain (NTD), a WD40 repeat domain, a helical bundle (HB) domain, and a C-terminal unstructured region (Figure 2A). The NTD and WD40 domains of Rrn6 reside in the palm where they engage Rrn11 (Figure 2D, Figure 2—figure supplement 1B), whereas the HB domain forms the knuckles and interacts with Rrn7 (Figure 2E). A flexible linker connects the WD40 and HB domains of Rrn6 (Figure 2A). No density was observed for the C-terminal region after H9 of the Rrn6-HB domain, in agreement with the lack of structure for this region (Knutson et al., 2014). Figure 2 with 2 supplements see all Download asset Open asset Core Factor architecture and Pol I interaction. (A-C) Ribbon diagrams showing the domain architecture of Rrn6 (A), Rrn11 (B), and Rrn7 (C). NTD, N-terminal domain; HB, helical bundle; CyclinC/N, C/N-terminal Cyclin Fold domain; IH, insertion helices; TPR, tetratricopeptide repeats. D and E, Rrn6's scaffolding role in Core Factor assembly by binding Rrn11 (D) and Rrn7 (E) using large interaction surfaces from WD40-NTD and HB, respectively. Color scheme is same as in (A-C). (F) Interface between Core Factor and the Pol I protrusion and subunit Rpb12. Overall views are also shown for each panel, with the same orientation and color scheme. The close-up view is indicated by a red box. Obstructing components are shown in transparency. https://doi.org/10.7554/eLife.27414.008 Rrn11 is predicted to contain a TPR (tetratricopeptide repeat) domain with N- and C-terminal unstructured regions (Knutson et al., 2014). In agreement with the prediction, we can assign most of Rrn11 to helical densities, with a total of 13 helices and three long loops (H2-H3, H4-H5, and H8-H9) (Figures 1B and 2B). The C-terminal helices from H5 to H13 resemble a classic TPR domain (Allan and Ratajczak, 2011; D'Andrea and Regan, 2003) more than H1-H4 (Figure 2B). Therefore, we named H5 to H13 the TPR domain of Rrn11, and H1-H4 the NTD. The NTD of Rrn11 resides in the palm (Figure 2—figure supplement 1A) and caps the WD40 domain of Rrn6, directly contacting repeats W3 to W5; the TPR domain interacts with repeat W6 and the NTD of Rrn6 (Figure 2D, Figure 2—figure supplement 1B) forming the thumb (Figure 2—figure supplement 1A). Like its counterpart TFIIB in the Pol II system, Rrn7 contains two cyclin fold domains (Knutson and Hahn, 2011; Naidu et al., 2011) (Figure 2C), which form the fingers (Figure 2—figure supplement 1A). The two cyclin fold domains can be aligned with those in TFIIB individually (Figure 2—figure supplement 2A), but a different relative orientation between them is adopted compared to their counterparts in TFIIB. One of the cyclin fold domains of TFIIB must be rotated when the other is aligned to its counterpart in Rrn7 (Figure 2—figure supplement 2B), to achieve the more compact organization in Rrn7. This compact architecture is presumably induced by Rrn6 intimately embracing Rrn7 (Figure 2E). The N-terminal cyclin fold (CyclinN) domain is mainly composed of 5 consecutive α-helices, whereas the C-terminal cyclin fold (CyclinC) domain contains a long insertion between H3 and H4. The insertion region is composed of 5 α-helices referred to as the Insertion Helices (IH) (Figure 2C). This finding is consistent with previous sequence analyses, in which the similarity of CyclinC with TFIIB stops at H3 (Knutson and Hahn, 2011). The helices H2-H4 of Rrn7 IH contacts Rrn11 TPR H5, H6 and H8, bridging the fingers with the thumb (Figure 2—figure supplement 1C). The interaction between Core Factor and Pol I in the Initial Transcribing Complex is mainly mediated by the Rrn11 TPR domain (the thumb) and the Pol I protrusion (Figure 2F). Compared to the large interfaces among Core Factor subunits, the interface between Rrn11 and the Pol I protrusion is rather small, involving Rrn11 helices H8, H10 and H12 (Figure 2F). Interestingly, Rrn7 IH H4 is also positioned near Rpb12 subunit of Pol I (Figure 2F), possibly contributing to Core Factor/Pol I interaction in the complex. The limited interface between Core Factor and Pol I is consistent with the flexibility of Core Factor observed in our density map (Figure 1—figure supplement 2C). Pol I and Core Factor are both involved in promoter DNA interactions The refined structure of the full complex clearly reveals the path of the promoter DNA in the Pol I Initial Transcribing Complex. Densities for both downstream and upstream duplex DNA were clearly resolved (Figure 1B). The downstream duplex DNA is inserted into the active site cleft of Pol I, stabilized by interactions with the clamp head, the cleft and the jaw domains of A190, the lobe domain of A135, and Rpb5 (Figure 3—figure supplement 1), similar to Pol II initiation complexes (He et al., 2016; Murakami et al., 2015; Plaschka et al., 2016) and Elongation Complexes of all three eukaryotic RNA polymerases (Barnes et al., 2015; Bernecky et al., 2016; Gnatt et al., 2001; Hoffmann et al., 2015; Neyer et al., 2016; Tafur et al., 2016). The positioning of the downstream duplex DNA upon promoter opening and during active elongation suggests conserved mechanisms of DNA translocation among all three eukaryotic RNA polymerases, consistent with the fact that all RNA polymerases share a conserved catalytic core complex (Vannini and Cramer, 2012). The resolution of our reconstruction of Core Factor at 4.2 Å hinders us from confidently resolving the register of the upstream promoter sequence. To we assembled the Pol Factor complex using a nucleic acid scaffold (Figure 3—figure supplement and obtained a cryo-EM reconstruction at an overall resolution of Å (Figure 3—figure supplement 2B). In this scaffold, the non-template was to −27 site as from the and the poly-T mismatch were also resulting in an duplex sequence −27 to in the upstream region (Figure 3—figure supplement 2A). In the we observed a clear of the duplex DNA density from the upstream compared to our reconstruction assembled on the full scaffold shown in Figure (Figure 3—figure supplement 2C). This density very well with a model of duplex DNA (Figure 3—figure supplement the two reconstructions we were to confidently assign the in the upstream of the upstream promoter interactions in the Pol I Initial Transcribing Complex reveals unexpected features that are distinct from Pol II initiation complexes (Figure 3). To the upstream DNA from during promoter a is induced in the Pol II Pre-Initiation Complex by the binding of and further stabilized the TFIIB cyclin recognition interaction the et al., and Hahn, et al., and 2000). In two consecutive of and near that are 5 are generated by contacts (Figure This is to the absence of TBP in our structure, because TBP does not rDNA promoter using its et al., 2004) and is TATA-box sequence within the Core Factor binding TBP is not in but rather Pol I transcription (Aprikian et al., 2000; et al., 2012; et al., The Core Factor-DNA interaction reported is more similar to the initiation factor to Core complex, which revealed a in the DNA at et al., Figure with 2 supplements see all Download asset Open asset Core Factor engagement with promoter DNA from −27 to −16. (A) Rrn7 and Rrn11, as well as the Pol I wall and protrusion domains with rDNA promoter, resulting in two consecutive of in promoter DNA 5 Rrn7, DNA binding domains and promoter DNA are shown as in their corresponding density for and surface for Rrn6 and the of Pol I are shown as in the (B) interaction with promoter DNA from to (C) rDNA promoter from to (D) of promoter DNA near by the wall and Overall views are also shown for each panel, with the same orientation and color scheme. The close-up view is indicated by a red box. Obstructing components are shown in transparency. The interactions between Core Factor and promoter DNA are mainly mediated by subunits Rrn7 and Rrn11 (Figure Rrn7 mainly its N-terminal cyclin fold domain to with promoter DNA from −27 to (Figure This interaction helices H3 and H5, as well as the loops and H4-H5, contacting mainly the of the promoter In addition to these we also observed the in the C-terminal cyclin fold domain into the of the DNA near the interface between Core Factor and DNA (Figure This is likely to be an because we failed to assemble the Pol Factor complex when we used a scaffold to when the in the non-template at −27 was Rrn11 contacts promoter DNA from to (Figure Like the interaction, Rrn11 mainly contacts DNA including its NTD helices and as well as H5 in the TPR domain (Figure The between helices and H9 of the Rrn11 TPR domain is between and promoter contacting the DNA near (Figure Core interfaces well with previous et al., 2004) and and Pol I also contacts the upstream DNA near (Figure this the duplex DNA is between the wall and the protrusion (Figure This is distinct from Pol II Pre-Initiation Complex, in which the upstream promoter DNA is positioned the Pol II cleft (He et al., 2016; Murakami et al., 2015; Plaschka et al., 2016). Interestingly, when Pol II the elongation the upstream duplex DNA Å (Barnes et al., 2015; Bernecky et al., 2016) to the same corresponding as the upstream DNA in the Pol I Initial Transcribing Complex. with the fact that upstream DNA is also stabilized by the protrusion domain in Pol I Elongation Complex et al., 2016), indicates that Pol I initiation complex is in an state by Core Factor at the initiation the Pol I Initial Transcribing Complex structure reveals a different of the promoter suggesting distinct initiation mechanisms between Pol I and functional states reveal conformational in Pol I Initial Transcribing Complex during transcription initiation To gain insight into the nature of Core Factor we performed 3D on three different classes that were produced by the 3D classification step of our data procedure (Materials and methods; Figure 1—figure supplement 2C). We obtained reconstructions of the Pol I Initial Transcribing Complex in three distinct functional states (Figure 4, Figure supplement 1), with overall of 4.2 Å, Å and Å for states we respectively 2, and of the three states reveals unexpected conformational with the of Core Factor (Figure 4). Figure 4 with supplements see all Download asset Open asset of structural states with key functional in Pol I Initial Transcribing Complex. (A-C) views showing the of the Core complex relative to Pol I in 1 (A), 2 (B), and (C). The represent the path of upstream DNA in 2, in which Core Factor to Pol red circles the of the Rrn7 N-terminal ZR and linker regions (B) and the A49 tandem winged helix and linker domains (C). of and for the Core complex by 1 and with 2 is also in A and Structural mobility of the active site cleft in 1 2 and The is shown as with the of the in where the clamp domain adopts the most contracted in D and the movement of the compared to The A49 linker and downstream duplex DNA are shown as density with ribbon in and respectively. in and represent the of the A49 linker and the non-template respectively. out view for is shown to the of Core Factor and upstream DNA adopt a of on Pol I, with a movement of up to (Figure 1 shows a that the upstream DNA and Core Factor are rotated more toward the of Pol I, the DNA from the active site (Figure Compared to the Core complex is rotated toward the front of Pol I in 2 (Figure the upstream promoter to into the active site cleft, whereas the Core complex resides at an in (Figure the between Core Factor and Pol I is small (Figure 2F), this of movement can be Indeed, of the Core Factor/Pol I interface shows that Core Factor and the upstream DNA around the where Rrn11 interacts with the Pol I protrusion domain (Figure supplement suggesting Rrn11 may be Indeed, previous has shown that of Rrn11 in a Core Factor can still as be for a Pol I interaction surface (Knutson et al., 2014). structural in both Core Factor and Pol I are stabilized in distinct functional In 2, the movement of Core correlates well with the stabilization of Rrn7 zinc on Pol I (Figure Figure supplement 1B). This likely the Rrn7 linker region to the template DNA in the cleft, suggesting that the Rrn7 zinc region a role similar to that of TFIIB during Pol II transcription Indeed, from the the zinc ribbon domain of Rrn7 failed to support the open complex formation (Figure supplement In addition, the Core Factor also failed to the in open complex formation in the whereas the complex did (Figure supplement This with the finding that Core Factor complex the Rrn7 can still be recruited to rDNA promoter (Knutson et al., 2014),