Out-of-plane motions in open sliding clamps: Molecular dynamics simulations of eukaryotic and archaeal proliferating cell nuclear antigen

Steven L. Kazmirski, Yanxiang Zhao, Gregory D. Bowman, Mike O’Donnell, and John Kuriyan*

*Department of Molecular and Cell Biology and Department of Chemistry, Howard Hughes Medical Institute, University of California, Berkeley, CA 94720; †Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720; and ¶Howard Hughes Medical Institute, The Rockefeller University, New York, NY 10021

Article's page in PNAS - Local copy

Abstract / Figures from the paper



Abstract

Sliding clamps are ring-like multimeric proteins that encircle duplex DNA and serve as mobile DNA-bound platforms that are essential for efficient DNA replication and repair. Sliding clamps are placed on DNA by clamp loader complexes, in which the clamp-interacting elements are organized in a right handed spiral assembly. In order to understand how the flat ring-like clamps might interact with the spiral interaction surface of the clamp loader complex we have performed molecular dynamics simulations of sliding clamps (proliferating cell nuclear antigen, PCNA, from the budding yeast, humans and an archaeal species) in which we have removed one of the three subunits so as to release the constraint of ring closure. The simulations reveal significant structural fluctuations corresponding to lateral opening and out-of-plane distortions of the clamp, which result principally from bending and twisting of the &beta-sheets that span the inter-molecular interfaces, with smaller but similar contributions from &beta-sheets that span the intra-molecular interfaces within each subunit. With the integrity of these &beta-sheets intact, the predominant fluctuations seen in the simulations are oscillations between lateral openings and right-handed spirals. The tendency for clamps to adopt a right-handed spiral conformation implies that once opened, the conformation of the clamp can easily match the spiraling of clamp loader subunits, a feature that is intrinsic to the recognition of DNA and subsequent hydrolysis of ATP by the clamp-bound clamp loader complex.



Figures


Figure 1. (A) Crystal structure of PCNA from Saccharomyces cerevisiae (Yeast; PDB code 1PLQ) (8). The structure consists of 3 subunits, each with 2 domains connected by a long linker. In the molecular dynamics simulation, one subunit was removed. The domains in the simulated structure will be referred to as domains I, II, III and IV. (B) A view of the &beta-sheet at the inter-molecular interface. The hydrogen bonds between the interfacial strands, indicated as dotted lines, are maintained during the simulations even when the structure forms a large right-handed spiral (see Suppl. Figure 6).




Figure 2. (A) rms deviation in Cα positions from the starting structure for the Yeast-I simulation. The large deviations over the entire dimer structure (blue) are a result of rigid body motions between the individual domains. The rms deviation in Cα atom positions for domain I of Yeast-I (magenta) is generally lower than ~1.5 Å suggesting that the structures are stable. (B) The rms deviation in the Cα positions for domain I compared to the starting structure (blue) when domain II is matched onto the starting structure. The average deviation is ~3 Å suggesting a relatively small rigid body motion of domain II with respect to domain I. A comparison to the crystal structure (gray) is shown in (C) for the structure at 7.3 ns. A larger deviation is observed at the intermolecular interface when measuring the rms deviation in Cα positions for domain III (magenta) when domain II is matched onto the starting structure. These larger deviations result from the deformation of the &beta-sheet at the intermolecular interface and allow for a large displacement of domain III (C), as shown for the structure at 7.8 ns. The graphs were window averaged over a 50 ps window.




Figure 3. (A) A trimer was constructed artificially from the simulated dimers and the shortest distance between any atoms in domain I and an artificial domain VI (see Figure 1) were calculated. (B) Using just the simulated dimer, domain IV was matched to the crystal structure and the displacement perpendicular to the ring formed by the starting structure was calculated for domain I. For 7 of the 9 simulations, the displacement was mainly in a positive direction, corresponding to a right hand spiral. (Suppl. Fig. 8B). For clarity the instantaneous displacements were smoothed by averaging over 50 ps.




Figure 4. (A) The two lowest frequency normal modes are displayed for the intermolecular &beta-sheet in the Yeast-I simulation. The lowest frequency mode involves a bending of the &beta-sheet while the next lowest frequency mode shows a twisting of the &beta-sheet. For each mode, the displacement of each Cα atom from the mean position is indicated by an arrow, the length of which is scaled by an arbitrary scale factor for clarity. Larger displacements are colored red and smaller ones blue. (B) The intermolecular &beta-sheet from the yeast crystal structure is shown in gray (1PLQ). When the molecular dynamics simulation at 7.8 ns is matched onto strands indicated by arrows, a significant right-handed twist to the sheet can be seen. A significant spiraling is observed at 7.8 ns suggesting that the twisting of the &beta-sheet is responsible. (C) In Yeast-III, a left-handed spiral was observed. This conformation is correlated with the tearing of the &beta-sheet in domain I (hydrogen bonds between residues 105 and 110 and residues 103 and 112 break, indicated by dotted lines in the stick representation of the backbone).




Figure 5. Schematic drawing showing how PCNA may open. In solution PCNA is normally in a closed ring form. Either spontaneously or through its initial interaction with RFC, the clamp opens. When domain I of the simulation is superimposed on the corresponding domain in the RFC-PCNA crystal structure the right-handed spiraling of the clamp follows the spiral of RFC and would allow for contact between PCNA and all five clamp loader subunits.



Supplemental Figure 6. The number of native intermolecular hydrogen bonds in the inter-molecular interface &beta-sheet. Over the course of all simulations, the dimers remain stable as the number of hydrogen bonds at the inter-molecular interface is fairly constant. Time courses are shown for simulation I (blue), II (magenta), and III (yellow) of each species. In the crystal structure, the number of native hydrogen bonds are 7, 8, and 5 for Yeast, Human and Archaeal PCNA, respectively. Note that the archaeal structure has fewer hydrogen bonds at the interface, which is thought to be compensated for by an increased number of interfacial ion pairs (11).



Supplemental Figure 7. The rms deviation for the Cα positions for the trimeric yeast PCNA simulation. Starting with the 1PLQ crystal structure, the entire ring was simulated. The rms deviation averaged ~1.7 Å suggesting a stable ring conformation with no major rigid body motions between domains.



Supplemental Figure 8. (A) A trimer was constructed from the simulated dimers and the shortest distance between any atoms in domain I and an artificial domain VI (see Figure 1) were calculated and plotted for the simulations. (B) Considering the dimer alone, domain IV was matched to the crystal structure and the average displacement perpendicular to the ring the starting structure was calculated for domain I. For 7 of the 9 simulations, the displacement was in a positive direction, affording a right hand spiral.



Supplemental Figure 9. The two lowest frequency normal modes are displayed for the intramolecular &beta-sheet in the Yeast-I simulation. The lowest frequency is a twisting of the &beta-sheet while the second lowest frequency shows a bending of the &beta-sheet.



Supplemental Figure 10. An overlay for Yeast II of the gap opening distance (blue) and the average Z displacement (magenta). In general, the point of highest Z displacement or 'spiraling' of the PCNA dimer lead to a reduction of the gap opening.



Supplemental Table 1