The structure of Pol I, which consists of 14 subunits with a total molecular weight of 590 kDa, has been previously described in atomic detail in an inactive conformation using X-ray crystallography^{5, 6, 7}. To determine the structure of transcribing Pol I, we performed single-particle cryo-electron microscopy (single-particle cryo-EM) with a reconstituted yeast Pol I elongation complex containing a DNA–RNA scaffold (Fig. 1a, Extended Data Fig. 1), similar to that used to study transcribing mammalian Pol II¹¹. Particle classification enabled us to reconstruct the Pol I elongation complex structure at 3.8 Å resolution from approximately 94,000 single particles (Fig. 1c, Extended Data Figs 2, 3). The electron density revealed the downstream DNA, the DNA–RNA hybrid (Fig. 1b), and all Pol I domains with the exception of the flexibly linked C-terminal domain of subunit A49 (refs 12, 13) and the C-terminal domain of subunit A12.2. An atomic model was obtained by fitting rigid domains of the Pol I crystal structure⁵, positioning nucleic acids from the bovine Pol II elongation complex structure¹¹, and manually rebuilding regions that were structurally altered (Extended Data Table 1).

a, Nucleic acid scaffold and interactions between Pol I and nucleic acid. Template DNA, non-template DNA and RNA are shown in medium blue, sky blue, and red, respectively. Filled circles represent nucleotides that were well resolved in the electron density. Pol I residues within 4 Å distance are depicted together with the subunit identifier (A for A190, B for A135). b, Electron density for the DNA–RNA hybrid with the final model superimposed. The active site metal ion A is depicted as a magenta sphere. c, Ribbon model of the Pol I elongation complex. The view is from the ‘front’¹⁴ with the incoming downstream DNA pointing towards the reader. The colouring of the surfaces is according to the standard polymerase subunit colouring: A190, grey; A135, wheat; A49, light blue; A43, slate; AC40, red; A34.5, pink; Rpb5, magenta; Rpb6, silver-blue; AC19, yellow; Rpb8, green; A14, hot pink; A12.2, orange; Rpb10, blue; Rpb12, lemon. Template DNA, non-template DNA and RNA are depicted in medium blue, sky blue and red, respectively.

• Full size image (688 KB)
• Download PowerPoint slide (905 KB)

Comparison of the resulting Pol I elongation complex structure with the previous Pol I structure^{5, 6} reveals that the active centre cleft is contracted by up to 13 Å (Fig. 2a). Contraction occurs through relative movement of the two major polymerase modules ‘core’ and ‘shelf’¹⁴, as predicted⁵. The shelf module moves together with the clamp domain as a single ‘shelf-clamp’ unit, slightly rotating with respect to the core module (Fig. 2e, Supplementary Video 1). Another module, the ‘jaw-lobe’, moves closer to downstream DNA by up to 7 Å (Fig. 2a, d). Comparison of the Pol I elongation complex with elongation complex structures of Pol II^{8, 9}, Pol III¹⁰, and bacterial RNA polymerase^{15, 16} reveal that all of these polymerases adopt a similar contracted conformation in their transcribing state and underscores the fundamental structural and mechanistic similarity of cellular RNA polymerases from bacteria to eukaryotes (Extended Data Fig. 4).

a, Comparison of structures of Pol I elongation complex (EC) (orange) and free dimeric Pol I (PDB: 4C2M; black) after superposition of their A135 subunits. Cleft width was measured between subunit A190 residue E414 and subunit A135 residue K434. For clarity, only subunits A190 and A135 are displayed. b, Electron density of the folded bridge helix in the Pol I elongation complex. c, Comparison of bridge helices in the elongation complex (orange) and free Pol I (black). d, Pol I elongation complex ribbon model coloured by four mobile modules. The peripheral subcomplexes A14–A43 and A49–A34.5 are omitted for clarity. e, Movements of polymerase modules upon cleft contraction. Ribbon models of free Pol I (grey) and elongation complex are shown after superposition of their core modules (omitted). Arrows indicate movement and rotation of the clamp-shelf and the jaw-lobe modules. Colour code as in Fig. 1.

• Full size image (564 KB)
• Download PowerPoint slide (818 KB)

In the elongation complex structure, the connector is detached from Pol I, as observed when Pol I is bound to the initiation factor Rrn3 (ref. 13, 17). The expander is also displaced, enabling Pol I to form extensive interactions with the DNA–RNA hybrid (Fig. 1a). The enzyme contacts the DNA template at positions +4 to −9 and the RNA transcript at positions −1 to −8 (+1 represents the nucleotide addition site). Pol I generally binds nucleic acids with the same elements as Pol II⁹, but uses several Pol-I-specific residues to contact the upstream part of the DNA–RNA hybrid. The active centre adopts a catalytically competent conformation. The bridge helix is folded throughout (Fig. 2b, c). The trigger loop has weaker electron density, indicating higher mobility. The tip of the trigger loop lacks density, allowing for binding of the nucleoside triphosphate substrate. The polymerase switch regions and cleft loops adopt similar positions as in the Pol II elongation complex⁹ except that fork loop 1 is bent away from the hybrid (Extended Data Fig. 5a), as in the Pol III elongation complex¹⁰ and in a Pol II initiation intermediate¹⁸.

The Pol I elongation complex structure also provides insights into the regulation of the intrinsic RNA cleavage activity of Pol I. RNA cleavage requires subunit A12.2 (refs 19, 20), which consists of two domains. The N-terminal domain resembles that of the Pol II subunit Rpb9, whereas the C-terminal domain corresponds to the catalytic domain of the Pol II RNA cleavage factor TFIIS^{21, 22}. In the elongation complex structure, the N-terminal domain of A12.2 remains at the outer rim of the Pol I funnel region, whereas its C-terminal domain is displaced from the pore that it occupies in the Pol I crystal structures^{5, 6, 17}. Displacement of the A12.2 C-terminal domain from the pore apparently occurs during cleft contraction because modelling of this domain in the pore results in a clash with the contracted shelf module (Extended Data Fig. 5b, c). Thus A12.2 can only enter the active centre when the cleft is fully or partially expanded. This predicts that Pol I adopts a partially expanded conformation during A12.2 action, which is required for RNA proofreading and polymerase reactivation after backtracking.

To investigate whether the structural differences between the Pol I elongation complex and the free Pol I dimer arise from nucleic acid binding or from conversion of a dimer to a monomer, we also solved the structure of monomeric Pol I in the absence of nucleic acids at 4.0 Å resolution using approximately 80,000 single particles from the same data set (Extended Data Figs 2, 6a, Methods). In this structure, the connector and expander were also displaced, but the cleft was only partially contracted, similarly to the Pol-I–Rrn3 complex^{13, 17} (Extended Data Fig. 6b, c). The central bridge helix remained partially unwound, and the C-terminal domain of A12.2 remained in the pore (Extended Data Fig. 6d–f), confirming that the partially expanded conformation is required for A12.2-dependent RNA cleavage.

Thus, conversion of the Pol I dimer to a monomer leads to a partially expanded conformation, but not to the fully contracted active conformation. The partially expanded conformation resembles the conformation observed when the enzyme adopts a paused²³ or an inhibited²⁴ state (Extended Data Fig. 6g). In both, the bacterial polymerase and Pol I, movement of a rigid shelf-clamp unit away from the core module allows for expansion of the cleft and a coordinated widening of the pore (called the ‘secondary channel’ in bacterial RNA polymerase). This movement involves a slight rotation of the shelf-clamp unit with respect to the core module, reflected in the term ‘ratcheting’ used in one study of the bacterial polymerase²⁴.

Available data thus suggest that RNA polymerases can adopt partially expanded and contracted conformations that are associated with inactive and active states, respectively. Binding of nucleic acids in the cleft apparently maintains the contracted conformation and excludes A12.2 from the pore, whereas rearrangements in the nucleic acids upon misincorporation or pausing could induce the partially expanded conformation that is transcriptionally inactive but enables A12.2 entry into the pore and enzyme reactivation by RNA cleavage. According to this model, transcription elongation can be regulated by allosteric coupling of nucleic acid binding with cleavage factor binding in the cleft and pore, respectively, via contraction and expansion of the polymerase.

To investigate the physiological relevance of the single-particle cryo-EM structure, we further determined the structure of the natural Pol I elongation complex that forms in yeast cells by promoter-dependent initiation on rDNA with the use of cryo-electron tomography (Fig. 3). We spread active rDNA genes from exponentially growing yeast cells onto an electron microscopy grid such that they formed ‘Miller trees’²⁵ (Extended Data Fig. 7). To overcome previous limitations in sample preparation, we used instant plunge-freezing to keep the sample in a close-to-native environment. The obtained images revealed the detailed arrangement of Pol I enzymes along rDNA, nascent RNA emerging from Pol I, and large densities at the RNA ends that resemble 5' classical knobs²⁶ (Fig. 3a, b). From the cryo-electron tomography images, we selected 11 complete Miller trees and several smaller Pol I trails, each containing 10–20 Pol I enzymes with associated RNA. This yielded 993 transcribing Pol I enzymes for further analysis.

a, 2-nm thick tomographic slice though a cryo-electron tomography image with two of the Miller trees, showing the terminal knobs (grey circles), the DNA (typical examples marked by blue arrows), the RNA (red-pink arrows), and the Pol I enzymes (yellow and dark yellow circles for Miller trees 1 and 2, respectively). Several nucleosomes are attached to DNA like beads on a string (white box). b, Three-dimensional surface rendering of Miller tree 1 in (a), showing the terminal knobs (light grey), DNA (blue), RNA (red), possible RNA-modifying complexes (cyan), and Pol I complexes (yellow). c, Schematic of three successive Pol I enzymes (of the upstream Pol I, the central Pol I, and the downstream Pol I) overlaid with their probability density localization (heat map) (Extended Data Fig. 8a, b). d, Fit of the Pol I elongation complex ribbon model from single-particle cryo-EM into the cryo-electron tomography reconstruction in grey. The good fit observed here is not possible with the expanded conformation of Pol I (Extended Data Fig. 9b). Colour code as in Fig. 1.

• Full size image (557 KB)
• Download PowerPoint slide (822 KB)

We observed that each rDNA gene is loaded with ~70 Pol I enzymes, which showed a median centre-to-centre distance of 18 ± 10 nm (Fig. 3c, Extended Data Fig. 8a), consistent with previous results²⁶. Only ~2% of the Pol I complexes were separated by a distance of less than 12 nm, which would allow for interaction between enzymes. Furthermore, consecutive enzymes show random relative orientations, arguing against specific interactions that were suggested previously²⁷. The incoming and exiting rDNA enclose an angle of ~150° measured at each triple of successive Pol I molecules (Fig. 3c, Extended Data Fig. 8b). This angle was independent of the length of the DNA between enzymes and could not be obtained from the single-particle cryo-EM structure, because density for upstream DNA was poor.

After classification we performed sub-tomogram averaging (n = 225) to obtain a cryo-electron tomography structure of the cellular transcribing Pol I at a resolution of ~29 Å (Fourier shell correlation (FSC) 0.5 criterion; ~25 Å with the FSC 0.143 criterion; Extended Data Fig. 9a). The cryo-electron tomography structure strongly resembled the single-particle cryo-EM structure, showing an overall cross-correlation score of 0.85. An FSC plot between the cryo-electron tomography and single-particle cryo-EM structures decreased beyond the 0.143 threshold at 31 Å (Extended Data Fig. 9a). The peripheral subcomplexes A14–A43 and A34.5–A49 were flexible, consistent with the weaker density observed in the single-particle cryo-EM structure (Extended Data Fig. 2). The width of the active centre cleft was the same in both structures (Fig. 3d), confirming that the contracted single-particle cryo-EM structure represented the natural conformation of actively transcribing Pol I (Extended Data Fig. 9b).

Taken together, we used here two independent cryo-electron microscopic approaches to define the contracted Pol I conformation as the active transcribing state of the enzyme. This revealed that all three eukaryotic RNA polymerases adopt a highly similar closed active centre conformation during transcription elongation. Together with published data, our results provide evidence that the elongation phase of transcription is regulated by cleft contraction and expansion. In particular, rearrangements of nucleic acids in the cleft above the active site cleft may be coupled to binding of factors in the pore beneath the cleft. Thus Pol I does not only undergo induced fit to align nucleic acids with the catalytic site, it is apparently also regulated allosterically.

Preparation of Pol I elongation complex (EC)

Endogenous 14-subunit Pol I was prepared from Saccharomyces cerevisiae as described previously⁵, with some modifications. Yeast strain CB010 expressing a C-terminal Flag/10× histidine-tagged A190 subunit was fermented and collected during the exponential phase. For Pol I purification, 350 g of cells were used. Proteins were precipitated overnight at 4 °C with ammonium sulphate (2 M). Re-solubilized Pol I was enriched by large-scale affinity purification with Ni-NTA beads (Qiagen). Further enrichment with anion and cation exchange chromatography yielded to pure Pol I enzyme. The sample was applied to a Superose 6 10/300 size-exclusion column (GE Healthcare) in 5 mM HEPES (pH 7.8), 150 mM potassium acetate, 1 mM MgCl₂, 10 μM ZnCl₂ and 10 μM β-mercaptoethanol.

DNA and RNA were purchased from IDT and Exiqon (Vedbaek), respectively. The nucleic acid scaffold sequences were as follows. Template DNA, 5′-AAGCTCAAGTACTTAAGCCTGGTCATTACTAGTACTGCC-3′; nontemplate DNA, 5′GGCAGTACTAGTAAACTAGTATTGAAAGTACTTGAGCTT-3′; RNA, 5′-UAUCUGCAUGUAGACCAGGC-3′ (in underlined nucleosides, a methylene bridge connects the 2′-O and the 4′-C atoms of the ribose ring, thereby forming locked nucleic acids). Nucleic acids were annealed by continuously decreasing temperature from 95 °C to room temperature over a period of 60 min. EC assembly was achieved by incubating Pol I (300 μg, 3.5 mg ml⁻¹) with a twofold molar excess of scaffold for 10 min at room temperature (Extended Data Fig. 1).

Single-particle cryo-EM

For single-particle cryo-EM, Pol I EC complexes at a concentration of 200 μg ml⁻¹ were cross-linked with 0.9 mM BS3 (Sigma Aldrich) for 30 min at 30 °C after optimization (Extended Data Fig. 1). The reaction was stopped by adding 50 mM ammonium bicarbonate, and the sample was purified by size-exclusion chromatography on a Superose 6 3.2/300 column (GE Healthcare) equilibrated in 5 mM HEPES (pH 7.8), 150 mM potassium acetate, 1 mM MgCl₂, 10 μM ZnCl₂ and 10 μM β-mercaptoethanol. A 4 μl aliquot of 100 μg ml⁻¹ purified sample was applied to a glow-discharged (10 s) R1.2/1.3 UltrAuFoil grid (Quantifoil), and plunge-frozen in liquid ethane (Vitrobot Mark IV (FEI) at 95% humidity, 4 °C, 8.5 s blotting time, blot force 14). Dose-fractionated movies (30 frames, 0.25 s each) were collected at a nominal magnification of 130,000× (1.05 Å per pixel) in nanoprobe energy-filtered transmission electron microscopy (EFTEM) mode at 300 kV with a Titan Krios (FEI) electron microscope using a GIF Quantum s.e. post-column energy filter in zero loss peak mode and a K2 Summit detector (Gatan). The camera was operated in dose-fractionation counting mode with a dose rate of ~7.5 electrons per pixel per second (0.25 s single frame exposure) and a total dose of ~56 electrons per Ų. Defocus values ranged from −0.6 to −3 μm with marginal (<0.1 μm) astigmatism. Global motion correction was performed as described²⁸, but single-particle cryo-EM images were not partitioned.

Single-particle cryo-EM image processing

Parameters of the contrast transfer function (CTF) on each micrograph were estimated with CTFFIND4 (ref. 29). In a first step, ~1,500 particles were picked with the semi-automated swarm method of EMAN2 e2boxer.py³⁰. Relion was used for the whole-image processing workflow³¹ unless stated otherwise. Reference-free 2D classes were generated, seven of which were used for template-based auto-picking after filtering to 20 Å. We extracted 401,000 particles from 2,300 micrographs with a 230 × 230 pixel box and used them for further processing. Pixels with 5 standard deviations from the mean value were replaced with random values from a Gaussian noise distribution. All images were normalized to make the average density of the background equal to zero during pre-processing. False-positive particles showing very bright dots, which were presumably gold contamination, were removed by manual inspection or unsupervised 2D classification. The remaining 282,000 particles were aligned on a reference generated from the PDB entry 4C2M⁵ filtered to 40 Å. To correct for local motion and for radiation damage, we used the movie processing function of Relion including ‘particle polishing’, in which the resolution-dependent decay caused by radiation damage is taken into account³¹. Local resolution was estimated as described^{32, 33}.

During classification of single-particle cryo-EM images (Extended Data Fig. 2), we first separated out particles lacking nucleic acids. To this end, the Pol I cleft of the average resulting after the first round of alignment was masked. The subsequent classification led to four classes: (1) nucleic acid-free Pol I (115,000 particles); (2) Pol I elongation complex (94,000 particles; hereafter referred to as ‘EC’); (3) Pol I elongation complex with an alternative DNA conformation (37,000 particles); and (4) other particles (35,000 particles). Among the nucleic acid-free polymerase particles, 80,000 particles displayed a defined position of the C-terminal domain of A12.2. We refer to the SP average of these particles as the ‘monomer’.

In a second step, a mask around the dimerization domain was applied to remove particles from which the A49–34.5 subcomplex dissociated. This led to 32,000 and 40,000 particles in case of the Pol I monomer and Pol I EC, respectively. To visualize the mobile stalk, we then applied a mask around A14/43 during refinement allowing only local searches.

Gold-standard Fourier shell correlations (FSCs) were calculated during the 3D refinement in Relion between two independently refined halves of the data. According to the FSC 0.143 criterion, global resolutions of 4.0 Å and 3.8 Å were estimated for Pol I monomer and EC structures, respectively, which were sharpened with temperature factors of −146 Ų and −149 Ų, respectively.

Structural modelling

Two separate models were built for the monomer and the Pol I EC. PDB entry 4C2M⁵ was used as the starting model in both cases. Models were constructed lacking the expander, connector and, in case of the EC, the C-terminal domain of A12.2. The models were further truncated by removing the peripheral subcomplexes A49–34.5 and A14–43. The starting models were placed in densities for the monomer and the EC by fitting in UCSF Chimera³⁴, followed by rigid body fitting with a Phenix real space refinement³⁵. Rigid body groups were defined based on module definitions originally proposed for Pol II¹⁴. A starting model for DNA and RNA was derived from bovine Pol II¹¹ and further refined. Structurally altered regions were adjusted to the density in COOT³⁶ followed by real space refinement in Phenix. To generate complete models, structure of subcomplexes A49–34.5 and A14–43 were fit into the classified map in Chimera. No changes were made within the domains during model building, except for A34.5 C-terminal tail. The models were validated using the FSC between the model and the map, EMRinger³⁷ and Molprobity³⁸.

Miller tree preparation and cryo-electron tomography imaging

Miller chromatin spreads²⁵ were prepared with some modifications as described³⁹, using the NOY1071 yeast strain with 25 copies of ribosomal DNA (rDNA) repeats⁴⁰. Yeast cells were grown to mid-log phase (absorbance (A₆₀₀) = 0.4) in YPG medium supplemented with 1 M sorbitol at 30 °C. YPG medium contains 1% (w/v) yeast extract, 2% (w/v) bacto-peptone and 2% glucose. 1 ml yeast cell culture in mid-log phase was added for 4.5 min to the preheated 20T zymolyase (Amsbio, Biotechnology) solution (5 mg/200 μl zymolyase in YPG medium at 30 °C) for a slight digestion of the yeast cell wall. Subsequently, the yeast cell culture was centrifuged at 13.000 rpm for 15 s and the pellet was resuspended in 1 ml of 0.0025% Trition-X-100 (Sigma-Aldrich) ddH₂O at pH 9.2 adjusted with pH 10 buffer (Thermo Fischer Scientific). The yeast suspension was transferred to a flask containing 5 ml of 11 mM KCl solution. The lysate was pipetted and incubated in a hydrophobic plastic Petri dish (Carl Roth GmbH + Co. Kg) placed on a shaker for 45 min. Sucrose was excluded from the sucrose-formalin solution as used in ref. 39. To fix chromatin, 400 μl of 37% formaldehyde (Sigma-Aldrich) solution was applied for 5 min. The yeast lysate was deposited on electron microscopy grids with a ~30 nm thick carbon support layer evaporated by a carbon coater 208Carbon (Cressington) and glow discharged for ~1 min using a home-made device. Subsequently they were placed within home-built grid chamber insets and centrifuged within an Eppendorf 5810R centrifuge (Eppendorf) for 5 min at ~2,200g at 20 °C. Before plunge-freezing the grids were transferred to an 11 mM KCl solution for which ddH₂O at pH 9.2 was used.

The grids were immediately plunge-frozen in liquid ethane by a Vitrobot Mark IV (FEI) with 25 blotting force, 3 s blotting and 10–15 s draining time and the blotting chamber set to 100% humidity at 10 °C. Cryo-grids were mounted into autoloader grids with C-clippings (FEI) in an EM FC6 cryo-microtome (Leica) that was cooled with liquid nitrogen under gaseous flow to −150 °C. During mounting, grids were visually inspected to determine whether they contained an intact carbon film.

Tilt-series were recorded using DigitalMicrograph (Gatan Inc.) at a nominal magnification of 33,000× (4.0 Å per pixel) in EFTEM mode at 300 keV using a Titan Krios with a GATAN GIF Quantum s.e. post-column energy filter in zero loss peak mode and a K2 Summit detector. The camera was operated in counting mode with a dose rate of ~15 electrons per pixel per second and a total dose of ~100 electrons per Ų. The tilt-series ranged from –63° to +63° with an angular increment of 2° and defocus set at −5 μm. Tilted images were fiducial-less aligned⁴¹ and reconstructed by super-sampling SART⁴². The CTF was measured and corrected in slices in 3D⁴³.

Reconstruction and segmentation of Miller trees

3D reconstructions were visualized with the EMpackage in Amira (FEI & Zuse Institute)⁴⁴ and analysed in TOM package⁴⁵. Segmentation of the Miller trees was performed manually in Amira by drawing contours encompassing individual features on mildly Gaussian low-pass filtered tomograms using the high-contrast option of super-sampling SART⁴².

Sub-tomogram averaging of Pol I enzymes

Sub-tomograms containing transcribing Pol I enzymes on rDNA were manually selected. The enzymes were re-centred using a Gaussian blob of the size of Pol I. The positions of all enzymes were subsequently indexed such that they were placed sequentially on the DNA. As the DNA was visible in the reconstructions, the indexing was unambiguous (Extended Data Fig. 8d). For sub-tomogram averaging (that is, the cryo-electron tomography structure) we selected five Miller trees according to the following criteria: (1) They visually showed a transcriptional directionality (several Miller trees were not completely visualized in the field of view). (2) All Pol I enzymes aligned according to the Miller tree directionality. (3) The RNA exit site matched previous observations¹¹ (Extended Data Fig. 8c, e). This resulted in a total of 225 Pol I enzymes contributing to the final sub-tomogram average.

Sub-tomogram averaging was then performed on each Miller tree individually. This was to guarantee that the directionalities of the enzymes were not mixed owing to the globular shape of the enzyme, the pseudo-symmetry axis, and the varying ice thickness of the recording area leading to different signal-to-noise ratio among the enzymes. The Euler angles were determined a priori for each of the three consecutive Pol I enzymes per Miller tree by calculating the vector from centre-to-centre position. Constrained sub-tomogram averaging was performed on sub-tomograms with 64 × 64 × 64 voxels using a spherical mask (~20 nm diameter). To ensure the robustness of the sub-tomogram averaging, two different starting references were used (1) the average of all rotationally pre-aligned Pol I enzymes per strand, and (2) a Gaussian blob of the size of Pol I. Both converged to approximately the same density. During sub-tomogram averaging of each individual Miller tree, polymerases were low-pass filtered and the alignment was run with a translational freedom of 10 voxels around the Gaussian blob refined position, a full rotational freedom for phi and psi, and a constrained rotational freedom of ±30 degrees for θ with 5 degrees sampling increment, until the average reached convergence. The missing wedge was taken into account during the entire alignment.

The sub-tomogram averages of each Miller tree were individually inspected and the orientation of the Pol I enzymes on each Miller tree was analysed. The 3′ to 5′ directionality of the enzymes on each Miller tree was analysed. If all enzymes had the same directionality (that is, the signal-to-noise ratio was sufficient to align them properly), their sub-tomogram average was used for further processing. If the enzymes had conflicting directionalities (including complete random directionality), their sub-tomogram average was rejected. Five Miller trees qualified for this criterion. Their enzyme directionality was visualized compared to the Miller-tree directionality, and they all conformed. Finally, 225 enzymes (from the 993 total enzymes in the tomograms) of the five selected Miller trees were subjected to a refined sub-tomogram averaging and the resulting cryo-electron tomography structure reached a resolution of ~29 Å with the FSC 0.5 threshold criterion (~31 Å when compared to the single-particle cryo-EM structure).

Additional cryo-electron tomography analysis

In the tomograms both the DNA and the RNA could be seen emanating from the enzymes (Extended Data Fig. 8c, d). They were manually localized as close as possible to the enzyme and subsequently sub-tomogram averaging was performed around this position. To obtain evidence for the RNA exit channel visualized in the single-particle cryo-EM structure, we made three independent attempts to manually select the position of exiting RNA on Pol I in the tomogram without prior knowledge of the structure (Extended Data Fig. 8e). The resulting point distribution of exiting RNA on the cryo-electron tomography structure agreed with the location of the RNA exit channel in the single-particle cryo-EM map and further confirmed the correct superposition of the two independent structures.

The distances of consecutive Pol I enzymes were calculated as the Euclidian distance between their centre-to-centre positions. For plotting the probability density function, one enzyme was centred, the downstream enzyme was placed on the x axis, and the upstream enzyme was placed on the plane. Between three consecutive neighbouring enzymes, the in-plane angle was estimated.

For fitting of structures to the cryo-electron tomography reconstruction, rigid body fitting of the cryo-electron tomography and the single-particle cryo-EM structures of the Pol I EC was performed automatically, using MATLAB scripts (MATLAB and Statistics Toolbox Release 2012b, The MathWorks, Inc., Natick), implemented in the TOM package⁴⁵ (all scripts are freely available upon request) as well as Chimera³⁴. This resulted in a global cross-correlation value of ~0.8 and a FSC shown in Extended Data Fig. 9a. The contour level for the cryo-electron tomography structure for volume rendering of our average was calculated from the theoretical molecular mass with an average protein density of 0.8 kDa nm⁻³.

Data availability statement

Cryo-electron microscopy densities were deposited in the Electron Microscopy Data Base under the accession codes EMD-4147 and EMD-4148 for the EC and the free monomer, respectively. Sub-tomogram average densities were deposited in the Electron Microscopy Data Base under the accession codes EMD-4149. Model coordinates were deposited in the Protein Data Bank under the accession codes 5M3F and 5M3M for the EC and the free monomer, respectively.

1. Roeder, R. G. & Rutter, W. J. Multiple forms of DNA-dependent RNA polymerase in eukaryotic organisms. Nature 224, 234–237 (1969)
   • CAS
   • OpenURL
   • ISI
   • PubMed
   • Article
2. Novello, F. & Stirpe, F. Simultaneous assay of RNA polymerase I and II in nuclei isolated from resting and growing rat liver with the use of alpha-amanitin. FEBS Lett . 8, 57–60 (1970)
3. Grummt, I. The nucleolus—guardian of cellular homeostasis and genome integrity. Chromosoma 122, 487–497 (2013)
   • CAS
   • OpenURL
   • PubMed
   • Article
4. Goodfellow, S. J. & Zomerdijk, J. C. Basic mechanisms in RNA polymerase I transcription of the ribosomal RNA genes. Subcell. Biochem . 61, 211–236 (2013)
   • CAS
   • PubMed
5. Engel, C., Sainsbury, S., Cheung, A. C., Kostrewa, D. & Cramer, P. RNA polymerase I structure and transcription regulation. Nature 502, 650–655 (2013)
   • CAS
   • OpenURL
   • ISI
   • PubMed
   • Article
6. Fernández-Tornero, C. et al. Crystal structure of the 14-subunit RNA polymerase I. Nature 502, 644–649 (2013)
   • OpenURL
   • Article
7. Kostrewa, D., Kuhn, C. D., Engel, C. & Cramer, P. An alternative RNA polymerase I structure reveals a dimer hinge. Acta Crystallogr. D Biol. Crystallogr . 71, 1850–1855 (2015)
   • CAS
   • OpenURL
   • PubMed
   • Article
8. Gnatt, A. L., Cramer, P., Fu, J., Bushnell, D. A. & Kornberg, R. D. Structural basis of transcription: an RNA polymerase II elongation complex at 3.3 A resolution. Science 292, 1876–1882 (2001)
   • CAS
   • OpenURL
   • ISI
   • PubMed
   • Article
9. Kettenberger, H., Armache, K. J. & Cramer, P. Complete RNA polymerase II elongation complex structure and its interactions with NTP and TFIIS. Mol. Cell 16, 955–965 (2004)
   • CAS
   • OpenURL
   • ISI
   • PubMed
   • Article
10. Hoffmann, N. A. et al. Molecular structures of unbound and transcribing RNA polymerase III. Nature 528, 231–236 (2015)
    • CAS
    • OpenURL
    • PubMed
    • Article
11. Bernecky, C., Herzog, F., Baumeister, W., Plitzko, J. M. & Cramer, P. Structure of transcribing mammalian RNA polymerase II. Nature 529, 551–554 (2016)
    • CAS
    • OpenURL
    • PubMed
    • Article
12. Jennebach, S., Herzog, F., Aebersold, R. & Cramer, P. Crosslinking-MS analysis reveals RNA polymerase I domain architecture and basis of rRNA cleavage. Nucleic Acids Res . 40, 5591–5601 (2012)
    • CAS
    • OpenURL
    • PubMed
    • Article
13. Pilsl, M. et al. Structure of the initiation-competent RNA polymerase I and its implication for transcription. Nat. Commun. 7, 12126 (2016)
14. Cramer, P., Bushnell, D. A. & Kornberg, R. D. Structural basis of transcription: RNA polymerase II at 2.8 angstrom resolution. Science 292, 1863–1876 (2001)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
15. Vassylyev, D. G., Vassylyeva, M. N., Perederina, A., Tahirov, T. H. & Artsimovitch, I. Structural basis for transcription elongation by bacterial RNA polymerase. Nature 448, 157–162 (2007)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
16. Sekine, S., Murayama, Y., Svetlov, V., Nudler, E. & Yokoyama, S. The ratcheted and ratchetable structural states of RNA polymerase underlie multiple transcriptional functions. Mol. Cell 57, 408–421 (2015)
17. Engel, C., Plitzko, J. & Cramer, P. RNA polymerase I-Rrn3 complex at 4.8 Å resolution. Nat. Commun. 7, 12129 (2016)
18. Plaschka, C. et al. Transcription initiation complex structures elucidate DNA opening. Nature 533, 353–358 (2016)
    • OpenURL
    • Article
19. Sosunov, V. et al. Unified two-metal mechanism of RNA synthesis and degradation by RNA polymerase. EMBO J . 22, 2234–2244 (2003)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
20. Kuhn, C. D. et al. Functional architecture of RNA polymerase I. Cell 131, 1260–1272 (2007)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
21. Kettenberger, H., Armache, K. J. & Cramer, P. Architecture of the RNA polymerase II-TFIIS complex and implications for mRNA cleavage. Cell 114, 347–357 (2003)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
22. Ruan, W., Lehmann, E., Thomm, M., Kostrewa, D. & Cramer, P. Evolution of two modes of intrinsic RNA polymerase transcript cleavage. J. Biol. Chem. 286, 18701–18707 (2011)
    • CAS
    • OpenURL
    • PubMed
    • Article
23. Weixlbaumer, A., Leon, K., Landick, R. & Darst, S. A. Structural basis of transcriptional pausing in bacteria. Cell 152, 431–441 (2013)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
24. Tagami, S. et al. Crystal structure of bacterial RNA polymerase bound with a transcription inhibitor protein. Nature 468, 978–982 (2010)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
25. Miller, O. L., Jr & Beatty, B. R. Visualization of nucleolar genes. Science 164, 955–957 (1969)
    • OpenURL
    • ISI
    • PubMed
    • Article
26. Osheim, Y. N. et al. Pre-18S ribosomal RNA is structurally compacted into the SSU processome prior to being cleaved from nascent transcripts in Saccharomyces cerevisiae. Mol. Cell 16, 943–954 (2004)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
27. Albert, B. et al. RNA polymerase I-specific subunits promote polymerase clustering to enhance the rRNA gene transcription cycle. J. Cell Biol. 192, 277–293 (2011)
    • CAS
    • OpenURL
    • PubMed
    • Article
28. Li, X. et al. Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM. Nat. Methods 10, 584–590 (2013)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
29. Rohou, A. & Grigorieff, N. CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192, 216–221 (2015)
    • OpenURL
    • PubMed
    • Article
30. Tang, G. et al. EMAN2: an extensible image processing suite for electron microscopy. J. Struct. Biol. 157, 38–46 (2007)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
31. Scheres, S. H. A Bayesian view on cryo-EM structure determination. J. Mol. Biol. 415, 406–418 (2012)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
32. Plaschka, C. et al. Architecture of the RNA polymerase II-Mediator core initiation complex. Nature 518, 376–380 (2015)
    • CAS
    • OpenURL
    • PubMed
    • Article
33. Cardone, G., Heymann, J. B. & Steven, A. C. One number does not fit all: mapping local variations in resolution in cryo-EM reconstructions. J. Struct. Biol. 184, 226–236 (2013)
    • OpenURL
    • ISI
    • PubMed
    • Article
34. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
35. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
36. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
37. Barad, B. A. et al. EMRinger: side chain-directed model and map validation for 3D cryo-electron microscopy. Nat. Methods 12, 943–946 (2015)
    • OpenURL
    • PubMed
    • Article
38. Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D 66, 12–21 (2010)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
39. Osheim, Y. N., French, S. L., Sikes, M. L. & Beyer, A. L. Electron microscope visualization of RNA transcription and processing in Saccharomyces cerevisiae by Miller chromatin spreading. Methods Mol. Biol . 464, 55–69 (2009)
    • PubMed
40. Machín, F. et al. Transcription of ribosomal genes can cause nondisjunction. J. Cell Biol. 173, 893–903 (2006)
    • CAS
    • OpenURL
    • ISI
    • PubMed
    • Article
41. Castaño-Díez, D., Al-Amoudi, A., Glynn, A. M., Seybert, A. & Frangakis, A. S. Fiducial-less alignment of cryo-sections. J. Struct. Biol. 159, 413–423 (2007)
    • OpenURL
    • PubMed
    • Article
42. Kunz, M. & Frangakis, A. S. Super-sampling SART with ordered subsets. J. Struct. Biol. 188, 107–115 (2014)
43. Kunz, M. & Frangakis, A. S. Three-dimensional CTF correction improves the resolution of electron tomograms. J. Struct. Biol. http://dx.doi.org/10.1016/j.jsb.2016.06.016 (2016)
44. Pruggnaller, S., Mayr, M. & Frangakis, A. S. A visualization and segmentation toolbox for electron microscopy. J. Struct. Biol. 164, 161–165 (2008)
    • OpenURL
    • ISI
    • PubMed
    • Article
45. Korinek, A., Beck, F., Baumeister, W., Nickell, S. & Plitzko, J. M. Computer controlled cryo-electron microscopy—TOM² a software package for high-throughput applications. J. Struct. Biol. 175, 394–405 (2011)
    • OpenURL
    • ISI
    • PubMed
    • Article

Download references

We thank O. Gadal and I. Lèger-Silvestre for technical assistance with the yeast Miller tree spreading technique, and H. Schwalbe for initial discussions. We thank T. Gubbey for initial experiments on the Pol I elongation complex and C. Bernecky, C. Plaschka and D. Tegunov for support with the single-particle data analysis. We thank T. Schulz for yeast fermentation. S.N. was supported by a PhD student fellowship from the Boehringer Ingelheim Fonds. P.C. was supported by the Deutsche Forschungsgemeinschaft (SFB860, SPP1935), the Advanced Grant ‘TRANSREGULON’ from the European Research Council (grant agreement No 693023), and the Volkswagen Foundation. A.S.F. was supported by the Deutsche Forschungsgemeinschaft (SFB 902) and the Starting Grant ‘JTOMO’ from the European Research Council.

Reviewer Information: Nature thanks R. Ebright, E. Nogales and E. Nudler for their contribution to the peer review of this work.