Continuous millisecond conformational cycle of a DEAH box helicase reveals control of domain motions by atomic-scale transitions

Abstract

Helicases are motor enzymes found in every living organism and viruses, where they maintain the stability of the genome and control against false recombination. The DEAH-box helicase Prp43 plays a crucial role in pre-mRNA splicing in unicellular organisms by translocating single-stranded RNA. The molecular mechanisms and conformational transitions of helicases are not understood at the atomic level. We present a complete conformational cycle of RNA translocation by Prp43 in atomic detail based on molecular dynamics simulations. To enable the sampling of such complex transition on the millisecond timescale, we combined two enhanced sampling techniques, namely simulated tempering and adaptive sampling guided by crystallographic data. During RNA translocation, the center-of-mass motions of the RecA-like domains followed the established inchworm model, whereas the domains crawled along the RNA in a caterpillar-like movement, suggesting an inchworm/caterpillar model. However, this crawling required a complex sequence of atomic-scale transitions involving the release of an arginine finger from the ATP pocket, stepping of the hook-loop and hook-turn motifs along the RNA backbone, and several others. These findings highlight that large-scale domain dynamics may be controlled by complex sequences of atomic-scale transitions.

Fig. 1. Prp43 in the closed state (PDB ID: 5LTA).

C-terminal domain (CTD, orange), RecA2 domain (cyan) and RecA1 domain (blue). Boxes show close-up views on the hook-loop (a), the hook-turn (b), and on the ATP-pocket (c).

Fig. 2. Domain movements during the conformational cycle.

a–c Opening transition and (d–f) closing transition of Prp43. a, d Front view and (b, e) top view at the beginning (gray) and end (multi-colored) of the respective process taken from the initial and final frames of the opening or closing trajectory, respectively. Arrows highlight the motions of RecA2 domain (cyan) and CTD domain (orange) relative to RecA1 domain (blue). c Center-of-mass distance between RecA1 and RecA2 domains during opening and (f) during closing. Dashed lines indicate the RecA1–RecA2 distances in the closed Prp43 structure (green, pdb code 5LTA) or in the open structure of the homologous Prp22 (blue, pdb code 6I3P). The time averages are drawn as dark red lines while the underlying data is drawn in pale red.

Fig. 3. Molecular switches of the opening process.

Conformational transitions of (a–c) the β-hairpin, (d–g) the arginine finger, (h–k) the hook-loop, and (l–o) the sensor loop during Prp43 opening. (a, d, h, l) Molecular switches in the initial and (b/f/i/m) final frames of adaptive sampling of the open process. (c, g, j, k, n) In the distance plots, the time averages are drawn as dark red lines while the underlying data is drawn in pale red. Critical atomic distances that quantify the progression of the opening transition, where dashed lines indicate the distances in the 5LTA and 6I3P crystal structures. a View from the backside of Prp43: K403 of the β-hairpin hydrogen bound to U4 in closed conformation and (b) shifted from U4 to U3 in open conformation. c K403–U3 distance versus cumulative simulation time. d Closed RecA1–RecA2 interface with the Arginine finger R435 hydrogen bound to T389 and located on top of the P-loop. e Broken R435–T389 hydrogen bond after 5 ns. f Final conformation with distant R435 and T389 residues and with the arginine finger located underneath the P-loop. g R435–T389 distance versus cumulative simulation time. h Closed conformation, hook-loop bound to U5 via G349 and T381. i Open conformation, G349 and T381 shifted from U5 to U4. S387 formed an H-bond to U5 similar to the conformation in panel M. j G349–U4 distance and (k) T381–U4 versus cumulative simulation time. l Serine finger S387 bent towards the ATP in closed conformation and (m) pointing towards the RNA forming an H-bond with U5. n S387–U5 distance versus cumulative simulation time. o S387–I383 distance vs. RecA1–RecA2 distance, revealing a marked correlation between the sensor serine conformation (helical or turn) and Prp43 opening. Supplementary Fig. S12 shows the relative position of the zoom-in features to the overall structure.

Fig. 4. PCA of isolated RecA1 and RecA2.

a Motion along the first PCA vector of the isolated RecA2 domain visualized as 6 frames from blue to orange. The hook-loop and sensor serine exhibit the largest contributions to RecA2 fluctuations after excluding the highly mobile β-hairpin from PCA. b Motion along the first PCA vector of the isolated RecA1 domain. The hook-turn largely contributes to RecA1 fluctuations. Movies of these movements are shown in Supplementary Movies S3 and S4.

Fig. 5. Molecular switches of the closing process.

Conformational transitions of (a–c) the arginine finger, (d–g) sensor serine, (h–l) RecA1–RNA interactions, and (m–p) β-hairpin during Prp43 closing. a, d, h, m Molecular switches in the initial and (b, e, i, n) final frames of adaptive sampling of the closing process. c, f, j–l, o, p In the distance plots, the time averages are drawn as dark red lines while the underlying data of each frame is drawn in pale red. Critical atomic distances that quantify the progression of the closing transition. Dashed lines indicate the distances in the 5LTA and 6I3P crystal structures. a Open RecA1/RecA2 interface with distant R435 and T389 residues. b Closed conformation with ATP in the binding pocket (stick representation), thereby bridging the closed RecA1/RecA2 interface. R435 is tightly bound to the phosphate groups of ATP. c R435/C_ζ–ATP/O3B distance, revealing an R435–ATP H-bond formation early during the closing process. d Open conformation with the sensor serine S387 in the helical state and forming an H-bond with U5 of the RNA. e Closed conformation with S387 in the loop conformation, bound to the ATP–Mg²⁺ complex, and reaching underneath the RecA1 R152–L167 helix. f R435–U5 distance during the closing process. g S387–I383 distance versus RecA1–RecA2 distance. The color indicates the cumulative simulation time. h Open conformation with R153 and R180 of RecA2 hydrogen bound to U6 and U7, respectively. i Closed conformation with R153 and R180 hydrogen bound to U5 and U6, respectively, shifted one nucleotide upstream relative to the open conformation. j–l R153–U5, R180–U6, and T195–U6 distances during the closing process. m Open conformation with N382 interacting with the β-hairpin. n Closed conformation with N382 interacting with R152, thereby connecting RecA2 and RecA1 domains. o, p R152–N382 distance and N382–β-hairpin distance. Supplementary Fig. S13 shows the relative position of the zoom-in features to the overall structure.

Fig. 6. Mechanism and models of RNA translocation.

a Schematic hypothesis of the complete translocation cycle. 1. Apo structure. 2. ATP-binding triggers opening of the CTD/RecA interface, allowing 3. binding of ssRNA and formation of the Protein–ATP–RNA complex, represented by PDB ID 5LTA. 4. ADP bound state after ATP hydrolysis phosphate release. 5. RNA-bound state with open RecA interface and weak RecA2–RNA contacts. 6. Stable open structure, represented by PDB ID 6I3P. 7. ATP-binding triggers closure of the RecA interface and sliding of RecA1 along the RNA. 8. Protein–ATP–RNA complex translocated by one nucleotide relative to the 3. step. 9. Transition back to the apo-structure after finalizing multiple RNA translocations. b The classical inchworm model. c Proposed inchworm/caterpillar model to illustrate both the center-of-mass motion of the RecA-like domains (dark blue and cyan) and the crawling of RecA-like along the RNA. Hydrogen bond partners of protein and RNA are sketched as caterpillar legs and as red lines, respectively. Since the RecA-like domains bind the RNA with four or five H-bonds in the closed and open state, respectively, the caterpillar requires five legs. The central leg, modeling the sensor serine S387, carries out a rotation to bind the free nucleotide binding site before reaching the open state.