Real-time capture of σN transcription initiation intermediates reveals mechanism of ATPase-driven activation by limited unfolding

Abstract

Bacterial σ factors bind RNA polymerase (E) to form holoenzyme (Eσ), conferring promoter specificity to E and playing a key role in transcription bubble formation. σ^N is unique among σ factors in its structure and functional mechanism, requiring activation by specialized AAA+ ATPases. Eσ^N forms an inactive promoter complex where the N-terminal σ^N region I (σ^N-RI) threads through a small DNA bubble. On the opposite side of the DNA, the ATPase engages σ^N-RI within the pore of its hexameric ring. Here, we perform kinetics-guided structural analysis of de novo formed Eσ^N initiation complexes and engineer a biochemical assay to measure ATPase-mediated σ^N-RI translocation during promoter melting. We show that the ATPase exerts mechanical action to translocate about 30 residues of σ^N-RI through the DNA bubble, disrupting inhibitory structures of σ^N to allow full transcription bubble formation. A local charge switch of σ^N-RI from positive to negative may help facilitate disengagement of the otherwise processive ATPase, allowing subsequent σ^N disentanglement from the DNA bubble.

Fig. 1. RPo formation kinetics of Eσ^N informs cryo-EM analysis.

a Stopped-flow measurement employs a linear, Cy3-labeled (yellow) σ^N promoter DNA (Aae dhsU). The fluorophore is subject to protein-induced fluorescent enhancement (PIFE) when RPo forms. For the experiment, an early-melted intermediate (box 1; E + σ^N+dhsU+C1) is pre-assembled from core RNAP (E, gray), σ^N (orange), and Cy3-labeled dhsU (green) in presence of bEBP (C1; blue). ATP is prepared in buffer (box 2). Mixing the contents of boxes 1 and 2 will start the reaction and PIFE can be observed. b RPo formation-dependent fluorescence saturates at around 8–10 min reaction time and is dependent on the presence of all components (black; n = 6). Mixing ATP with complexes without C1 (red; n = 5), ATP with DNA alone (blue; n = 3), or the E + σ^N+dhsU+C1 with buffer (no ATP, green; n = 6) results in only background signal. Solid lines represent the average of the individual mixing events. Dotted lines indicate error bands of one standard deviation. The arrow indicates the time point used for cryo-EM sample preparation. c Cryo-EM analysis of samples prepared after 35 s reaction time yields three major populations: early-melted intermediates (RPem, 31% of particles), bEBP-bound complexes (55% of particles), and complexes with a melted promoter DNA (RPo-like, 14%). Maps from the 3D classification (see main text and methods for details) were colored according to core RNAP (gray), σ^N (orange), dhsU (green), and bEBP (blue). The dashed arrows indicate the sequence of the complexes along the transcription initiation pathway. Source data are provided as a Source Data file.

Fig. 2. Cryo-EM structures capture translocation and remodeling of σ^N-RI by the bEBP.

a Overview of the merged, unsharpened cryo-EM map of the “open” ring state. Map density is colored according to the location of subunits, non-template (nt) and template (t) strands. The extent of visible map density for the DNA duplex is indicated by the colored letters at the top. b Helix 1 (H1) of σ^N-RI exhibits three turns in RPem (left) and RPem-bEBP^ADP-AlFx (middle), but a full turn is unfolded in the RPi1 intermediate (right). Residue P17 (red) located at the N-terminal end of H1 in RPem serves as visual indicator of helix changes. Regulatory interactions between σ^N-RI and the ELH remain intact (dotted circle; residues in magenta are mutation sites for bypass variants). Bases corresponding to the −11 position in dhsU are colored in yellow. c Comparison of the σ^N-RI:bEBP interaction between RPem- bEBP^ADP-AlFx (left) and RPi1 (right), illustrating the translocation of σ^N-RI by two residues. Every second residue in σ^N-RI starting with M1 up to Q11 is highlighted in red. M1 was not modeled for the open and closed ring structures.

Fig. 3. An engineered proteolysis assay measures translocation by the bEBP.

a The structures suggest that the bEBP (blue; shown with C-terminal domains) actively extrudes σ^N-RI (orange) from the initiation complex (left panel). The mycobacterial proteasome requires an ATPase (Mpa, purple) to feed protein substrates (green) into its proteolytic chamber (middle panel). A short C-terminal motif (GQYL) maintains interaction between proteasome and ATPase (inset). In our assay, an engineered variant of the bEBP carrying the GQYL motif (light blue) interacts with the proteasome. Proteolysis of σ^N-RI provides a measure for bEBP-mediated translocation during transcription initiation. b Schematic representation of the variants used for the proteolytic assay. c Distinct band shifts of 2Nt-σ^N, GS7-σ^N, and Nt-σ^N in presence of ATP and the bEBP variant carrying the GQYL motif (C1-GS4) demonstrate translocation of σ^N-RI into the proteasome by the bEBP. Reactions contained 0.5 μM RNAP, 0.5 μM σ^N, 0.6 μM dhsU-CT, and 0.5 μM C1 or C1-GS4 and 5 mM ATP or ATPγS as indicated. Representative gels of at least two individual experiments are shown. d Processive translocation by the bEBP occurs with σ^N, C1-GS4, and the proteasome. A translocation halt is only observed with the full transcription complex. Representative gels of three individual experiments are shown. e Estimated length required to reach the proteasome active site predicts bEBP position on σ^N-RI during translocation halt (black shading). Secondary structure elements of σ^N-RI present in RPem (H1 and H2) are indicated as gray dashed boxes. The experimentally determined cleavage site is indicated in red (*) on the sequence of Nt-σ^N and the σ^N-AAA variant. Conserved net charge distribution of σ^N-RI (solid line) changes from positive to negative around residues 28–30. Charge distribution of σ^N-AAA (dashed line) shown for comparison. Sequence logo is based on the alignment shown in Supplementary Fig. 9. f The surface of the inner pore of the bEBP is lined with negative charges near the hydrophobic pockets that capture residues during translocation. Solvent-excluded surface was clipped to reveal the pore interior and colored according to Coulombic electrostatic potential calculated in ChimeraX. Source data are provided as a Source Data file.

Fig. 4. Cryo-EM structures capture intermediate states of bEBP ATP hydrolysis.

a Conserved “GAFTGA” loops at the bEBP pore (top panel) engage with σ^N-RI and arrange in a spiral pattern (bottom panel; residues 209–224 of each bEBP subunit and residues 3–15 of σ^N-RI are shown). Map density for σ^N-RI (sharpened map) shown as transparent surface (orange). The small inset shows a schematic representation. b Clipped surface representation of bEBP subunits reveals captured side chains of σ^N-RI in hydrophobic pockets. Pocket-forming residues are shown in stick representation. Subunits a and f are hidden for visual clarity. c Open and closed ring states correspond to ATP- and ADP-bound states of subunit e. ATP colored in red, ADP colored in gray. The C-terminal domain of subunit e extends below the plane of the hexamer in the ATP-bound state (right), but swings into the plane of the ring in the ADP-bound state (left). Small insets indicate conformational change of subunit f schematically. d Active site interactions in subunit e show ATP coordination in a pre-catalytic state (left panel; open ring). Closed ring state contains ADP in the active site of subunit e, corresponding to a post-catalytic state. The arginine finger (R299, f) moved out of the active site (black arrow), and intersubunit salt bridge between E174 (e) and R253 (f) is broken. S-I = sensor I; S-II = sensor II.

Fig. 5. Stabilization of the DNA t-strand by σ^N-RII in RPo and RPo+2A.

a Unsharpened map of the RPo complex containing a transcription bubble opened from −9 to −1 (top schematic). b Unsharpened map of RPo+2 A, an initial transcribing complex containing nucleotides (red) in the active site. The transcription bubble is opened from −9 to +3 (top schematic). c Schematic comparison of melted regions in the DNA for complexes along the σ^N transcription initiation pathway. d In RPo (and RPo+2A), σ^N-RII (red) loops around the DNA t-strand (pale green; left panel) and exits the complex between β-Si2/flap domain and the β protrusion (right panel). The model of RPo is shown with β/β’ subunits in clipped surface representation to reveal the interior. Labeled arrows indicate neighboring structural elements in σ^N for clarification. e Active site of RPo+2A contains ADP in the i site and ATP in the i + 1 site in a pre-catalytic state. The trigger helix (blue) coordinates triphosphates of ATP (i + 1). Density of the sharpened map around DNA and key residues shown as transparent surface.

Fig. 6. Schematic of the proposed mechanism.

a For ATP hydrolysis-coupled substrate translocation by the bEBP, the ATP-containing bEBP binds to an unstructured end of the protein substrate (σ^N-RI) to engage in a spiral conformation, leaving a gap between the top (blue) and bottom (yellow) subunit. ATP hydrolysis in the green subunit disrupts the intersubunit interaction between green and yellow allowing the yellow subunit to reach across the gap and contact the blue/top subunit. The pore loop of the yellow subunit engages with the substrate, and thereby, the yellow subunit assumes the position at the top of the spiral and the complex advanced one step along the substrate. Subsequently, ADP in the green subunit is exchanged for ATP and cycle can start over. b For the isomerization from RPi1 to RPo, the transcription initiation complex is viewed along the DNA axis and large subunits (β, β’, bEBP) are clipped along the axis of σ^N-RI (red) for visual clarity. During translocation of σ^N-RI by the bEBP, the ELH must insert at the minor groove near the DNA bubble to place the DNA t-strand between σ^N-RI and the ELH. To loop around the t-strand, σ^N-RI/RII passes around the ELH and inserts between β-Si2 and the β protrusion forming the arrangement observed in the RPo and RP+2A structures.