Previous Article |
Table of Contents
| Next Article 
BIOCHEMISTRY
Open clamp structure in the clamp-loading complex visualized by electron microscopic image analysis
*Department of Structural Biology, Biomolecular Engineering Research Institute, 6-2-3 Furuedai, Suita, Osaka 565-0874, Japan; and Laboratory of Protein Chemistry and Engineering, Faculty of Agriculture, Kyushu University, Hakozaki, Higashi-ku, Fukuoka-shi, Fukuoka 812-8581, Japan
Communicated by John Kuriyan, University of California, Berkeley, CA, July 28, 2005 (received for review June 13, 2005)
 |
Abstract |
|---|
 Top Abstract MethodsResults and DiscussionReferences |
|---|
AAA+ ATPase | clamp loader | DNA replication | electron microscopy | single-particle analysis
We previously reported the 23-Å resolution EM structure of a clamp-loading RFC–PCNA–DNA ternary complex from Pyrococcus furiosus (Pfu), which was stabilized by introducing a nonhydrolyzable ATP analog, ATP
S (17). The structure showed the two building blocks, a larger horseshoe and a smaller closed ring. It appeared the best interpretation based on the available data that the horseshoe and the closed ring correspond to RFC and PCNA, respectively. Although the atomic structures of the PCNA trimer (18) and RFC small subunits (RFCSs) (11) were available, along with the information about the 1:4 stoichiometry for RFC large subunit (RFCL) and RFCS in the RFC hetero-pentamer (5), the fitting of the atomic model into the EM map was not completely satisfactory, and some ambiguity remained. In this study, we describe the EM structure of the clamp-loading complex determined at 12-Å resolution. The more excellent fitting of the atomic structures of the RFC and PCNA subunits into the map allows us to ensure that this is a real 3D structure. We now see that the PCNA is kept open in the complex with a spring-washer-shaped conformation. The previous EM map was conversely assigned, with a closed PCNA and an open C-terminal collar of RFC. The first direct view of the open clamp structure provides an opening mode of the clamp during the clamp-loading reaction.
|
Methods |
|---|
|
TopAbstractMethodsResults and DiscussionReferences |
|---|
PfuPCNA and PfuRFC were mixed with either of the priDNAs and ATPS and incubated at 37°C for 30 min. The mixture was repurified with a Superdex 200 gel filtration column. The fractions corresponding to the reconstructed complex were applied onto the carbon-coated grids, stained with 2% uranyl acetate, and examined with a JEOL JEM-100CX electron microscope equipped with a LaB6 electron gun. The images obtained on Fuji FG films were digitized with a DSScanner 1010.
Single-Particle Reconstruction of the Nonlabeled Complex. Three data sets of the nonlabeled clamp-loading complexes were processed as follows. All of the processing for image analysis was carried out with the EMAN software suite (19). Particle images were picked up semiautomatically by the program BOXER. Estimation of the contrast transfer function (CTF) parameters and correction of the CTF phase flipping were performed by using the CTFIT program. The 3D models were generated by using the STARTNRCLASSES and STARTANY programs and were used as the initial models for the low-resolution refinement process. In this refinement process, the low-pass filtered particle images were used, and no CTF correction was performed. Because the resolution and the convergence in the refinement of the pri11/30 data set were much better than those of the others, only the pri11/30 data set was applied to the subsequent high-resolution refinement processes, including CTF amplitude correction, automatic 3D masking (20), and Fourier ring correlation, instead of cross-correlation factor for the criterion of image similarity. The total iteration number of refinements was 100. The number of particle images used for the final 3D reconstructions was 19,495. The resolution estimated by the 0.5 criterion of the Fourier shell correlation was 12 Å.
|
EM experiments, film scanning, and particle picking-up were carried out as described above. The refined 3D map of the nonlabeled complex was used for the initial models for both data sets, and the low-resolution refinement processes were performed. The additional density for the labeled streptavidin was poorly visualized in the refined 3D maps, indicating that the structures of the labeled complexes were heterogeneous, presumably because of the variability of the streptavidin-labeled DNA termini or from contamination of the labeled and nonlabeled particles in each data set.
Therefore, further refinement was performed by using the MULTIREFINE program (21), in which multiple 3D models are refined simultaneously by using a single data set. For each of the data sets, two 3D maps, model A and model B, were prepared as the initial models for the labeled and nonlabeled complexes, respectively. As model A of each data set, the refined 3D maps obtained from the prior conventional refinement were used. On the other hand, the low-pass filtered maps of the nonlabeled complexes were used as model B. The numbers of particle images used for the final 3D reconstructions and the estimated resolutions were as follows: 3,583 particles and 22 Å for class A of the ds-labeled (3D model not shown), 3,718 particles and 22 Å for class B of the ds-labeled (yellow in Fig. 1d), 2,568 particles and 23 Å for class A of the ss-labeled (magenta in Fig. 1d), and 2,365 particles and 23 Å for class B of the ss-labeled (3D model not shown). To identify label positions on the complex, automatic fitting with the FOLDHUNTER program (22) was performed to fit the 3D maps of labeled complex into that of the nonlabeled complex.
Model Fitting and Graphical Representation. All of the 3D graphical representations were produced by using the University of California, San Francisco CHIMERA package (23). The volume for the 3D surface rendering of the nonlabeled complex was estimated from the theoretical molecular mass of the clamp-loading complex, 302.4 kDa, and an average protein density of 0.81 Da/A3. The volumes of the labeled complexes were chosen to create surface shapes similar to that of the nonlabeled complex.
The atomic coordinates of proteins, PfuRFCS chain B in the hexameric form [Protein Data Bank (PDB) ID code 1IQPB] and the PfuPCNA trimer (PDB ID code 1GE8 [PDB] ), were interactively fitted as rigid bodies by using the 3D viewing mode of CHIMERA. An ideal B-form DNA with 11 base pairs (ATATATATATA and TATATATATAT) was fitted by using the QSEGMENT program, which is an interacting volume-segmentation tool, and the FOLDHUNTER program.
To calculate the difference 3D map between the EM 3D map and the atomic model, the density value of the EM map and the 12-Å-resolution map of the atomic model, which was generated by the PDB2MRC program, were normalized to make their means of density 0 and their standard deviations 1. The difference map calculation was performed by the PROC3D program of EMAN.
The simplified 3D structures shown in Fig. 4 were created by using the Volume Path Tracer tool in CHIMERA. In the 3D space, the "markers" were placed interactively onto the representative vertexes of every domain of the atomic coordinates. Appropriate markers were linked to each other with the sticks.
|
Results and Discussion |
|---|
|
TopAbstractMethodsResults and DiscussionReferences |
|---|
(9) and yeast RFC (12) clamp loader complexes. Consequently, the remaining open ring in the middle of the complex should be assigned to the N-terminal AAA+ domains of RFC. Determination of the priDNA Termini in the Clamp-Loading Complex by the Streptavidin Labeling. In our new map, a rod-like density, roughly ranging over a 10-nt dsDNA, is encircled by the lower six-knotted open ring, suggesting that even the DNA can be partly visualized (Fig. 1c, front and bottom). In fact, further DNA labeling experiments, using streptavidin (24), revealed the functional binding of DNA. For selective labeling, the clamp-loading complexes were reconstituted, using a 3' recessed primed DNA consisting of a 25-nt primer and a 40-nt template strand (pri25/40), with a biotinylated 5' end on either the primer (Fig. 1b, yellow) or template strand (Fig. 1b, magenta), which was subsequently bound to streptavidin. For the 3D structure calculation of these labeled complex, the multireference refinement gave the improved 3D structures (see Methods for details). In the ss-labeled complex, the class A structure showed a clear globular density for the streptavidin, whereas the class B did not. This result indicates that the refinement successfully separated the labeled complexes from the nonlabeled ones. In the refinement for the ds-labeled complex, unexpectedly, the additional density was seen in the class B structure, whose initial model was derived from the nonlabeled complex, whereas the class A structure showed a deformed morphology. Thus, the ds-labeled complexes, with the structure similar to the nonlabeled ones, were classified into class B, and exhibited a clear globular density for streptavidin, whether or not the initial model holds additional density. The other particles grouped into class A could be heterogeneous, and thus resulted in a deformed structure.
In the 3D structure of the streptavidin-labeled complex with the biotinylated 5' end of the primer strand (ds-labeled complex), an additional globular density, assigned to the bound streptavidin, was observed below the rod-shaped density within the lower open ring (yellow arrows in Fig. 1d). On the other hand, in the other complex containing pri25/40 biotinylated at the 5' end of the template strand (ss-labeled complex), the same additional density was seen near the crack of the middle open ring (magenta arrows in Fig. 1d). These results strongly suggest that the DNA approaches from the open ring side and runs toward the central chamber of RFC. The 5' overhang of the template strand passes through the crack of the N-terminal domains of RFC and is positioned outside of the complex, whereas the recessed 3' end of the primer is located within the center of RFC (Fig. 4). Therefore, the rod-shaped density within the open PCNA should correspond to the ds region of the priDNA. This feature completely supports the DNA binding mode proposed from the yeast RFC–PCNA complex (12).
Fitting of the PfuPCNA and PfuRFCS Atomic Structure. The fitting of the atomic structures of the RFCS and PCNA subunits into the density allowed us to build a reasonable model (Fig. 2 and Movie 1, which is published as supporting information on the PNAS web site). First, the three PCNA subunits were fitted into the six-knotted open ring, which holds the interface with other proteins facing to RFC. Then, the five RFCSs were fitted into the map, so that the C-terminal collar domains can be located in the five-knotted closed ring and the N-terminal AAA+ domains in the middle region. Although we did not perform any domain rearrangement of the subunits or molecular dynamics simulation, iteration of manual rearrangements allowed the building of a suitable atomic model without any serious steric clash between the subunits. Using the FOLDHUNTER utility (22) in EMAN, we could also consistently place an 11-nt dsDNA model into the rod-shaped density, which was encircled by the lower open ring. This remarkably good fitting, which highlights the accuracy of the density assignment, convinced us that the present EM map represents the true 3D view of the archaeal RFC–PCNA–DNA ternary complex, and proved that our careful and unique approach by the combination of useful techniques for 3D reconstruction, such as image comparison by Fourier ring correlation and automatic 3D masking in the refinement process (see Methods), is effective even on the negatively stained images.
|
complex alone (Fig. 3a) (9), rather than the structure of yeast RFC complexed with PCNA (12). On the other hand, the comparison of our EM structure with the yeast RFC–PCNA complex crystal structure revealed that the archaeal and yeast clamp loaders adopt a common domain organization, including the similar shapes of the archaeal RFCL and the yeast RFC1 subunit. In our atomic model of the ternary complex, one RFCS at the edge of RFC should correspond to RFCL, considering the positions of the functionally equivalent E. coli 
(9) and yeast RFC1 (12) subunits in their clamp loader complexes (Fig. 2a, gold subunit, or Fig. 3c, yellow subunit). Consistent with this prediction, we found an extra density near the crack of the N-terminal domain of RFC, in a difference density map that was calculated between the EM experimental map and a map computed from the atomic structures of the PCNA and RFCSs and the DNA duplex (Fig. 2b, magenta density). This density lies at a position equivalent to that occupied by the C-terminal domain IV of yeast RFC1 (12), suggesting that it could be assigned to the C-terminal domain of RFCL and could possibly contact PCNA (Fig. 3 b and c). Thus, we believe that archaeal RFCL and yeast RFC1 share a unique extra C-terminal domain, in addition to the N-terminal crescent, which consists of three common domains in all of the clamp loader subunits. Open Clamp Structure in the Clamp-Loading Complex. The PCNA structure within the clamp-loading complex is quite remarkable, because this presents the first direct view of the open state of the clamp ring. Intriguingly, the intrinsic planarity of the PCNA trimer, observed in the crystal structure (18), is deformed to the shape of a spring washer-shaped ring (Fig. 2a). This spiral configuration allows PCNA to contact RFC through a broader interface than that in the yeast RFC–PCNA binary complex (12): The contact in the archaeal complex could involve all of the five RFC subunits, in contrast to the contribution of only three RFC subunits to the interface in the yeast complex. This RFC–PCNA interaction mode, which enables PCNA to broadly contact the N-terminal surface of RFC but cannot be achieved by its planer conformation, may be a major factor for PCNA ring opening. Notably, the spiral pitch for the PCNA and the N-terminal domains of RFC is roughly one-half (about 3 Å translation/60° rotation) of that for the arrangement of the RFC AAA+ domains in the yeast RFC–PCNA complex (about 6 Å translation/60° rotation) (12). Molecular dynamics simulation indicated that archaeal and yeast PCNA could move into a spiral structure similar to that in our clamp-loading complex (25). Even taking the 12-Å resolution limit into account, the 5-Å gap size of the open PCNA appears to be too small for at least dsDNA, and possibly for ssDNA, to enter it. Therefore, there could exist other states of the ternary complexes in which PCNA opens more widely. On the other hand, we cannot exclude a possibility that PCNA first pass the ssDNA and then slides on DNA to the primed position.
|
clamp loader alone (9) would correspond to a state before the beginning of the clamp-loading reaction (Fig. 4a), whereas the yeast RFC–PCNA binary complex (12) may represent a view just after the PCNA closed onto the DNA (Fig. 4e). On the other hand, it is likely that our present RFC–PCNA–DNA ternary complex structure corresponds to an intermediate state between the above two structures (Fig. 4c).
|
complex alone (Fig. 4 a and b). However, in general, ATP binding to a clamp loader is essential for its functional interaction with a sliding clamp, and therefore the ATP binding should cause a structural change, even though it is small. To define such a small but essential structural change of the clamp loader along with the clamp-loading reactions, further precise structural studies are required. Notably, the binding of two ATP molecules to the E. coli complex has triggered no significant structural change from the nucleotide-free "inactive" complex, suggesting that the binding of the third ATP (the complex binds up to three ATPs) could cause a structural change of the clamp loader to an active form (27).
Secondly, PCNA remains open in our structure, where the ATP has not been hydrolyzed yet, suggesting that the detachment of RFC from the ternary complex is strongly coupled with ATP hydrolysis by RFC. Notably, both the yeast binary complex and our ternary complexes lie in ATP-bound states, while they assume substantially different PCNA and RFC conformations. At the moment, the mechanism of their conformational conversion remains elusive. One possible explanation is that the RFC open–PCNA open state (our archaeal complex, Fig. 4 c and d) could be transformed to the RFC closed–PCNA closed state (the yeast complex, Fig. 4 e and f) by the consumption of very little energy. Our EM view and the yeast RFC–PCNA complex structure may reflect distinct scenes, which happened to selectively trap various nucleotide-bound states, depending on the sample preparation and/or mutations in the arginine finger of the RFC subunits. It may also be possible that the conversion between the two states depends on the stronger affinity of RFC for DNA than for PCNA, while the functional conversion appears to be coupled with the hydrolysis of some bound ATP molecules and possibly with the recognition of the 3' ends of the primer strands by RFC. For instance, upon binding to PCNA, RFC could convert the closed PCNA ring to an open form, while RFC retains essentially the same conformation. Then, the interaction of RFC with a primed DNA may induce its conformational change to adjust to the DNA spiral. In the yeast model, 
10 base pairs of the DNA duplex could be involved in the interaction with the RFC subunits. On the other hand, in our present model, RFC appears to contact four base pairs or less on its central chamber. From the present 3D map, determined at a medium resolution, it is uncertain whether RFC contacts the 3' end of the primer, although in the current atomic model the dsDNA region may be slightly short for RFC to interact with the 3' end of the primer. Subsequently, PCNA may reduce its extensive contact with RFC and regain its planer ring form, thereby completing the topological link to DNA (Movie 2, which is published as supporting information on the PNAS web site). In the entire course of this reaction, the C-terminal collar region of RFC could retain its firm pentameric ring structural base, to allow the clamp-loading reaction to proceed smoothly without RFC subunit separation.
The crucial, but elusive, question is the true role of ATP hydrolysis during the clamp-loading event. At the moment, it is not possible for us to propose a unified model that can account for all of the reported data about ATP hydrolysis. Further progress will be required for a more precise understanding of the clamp-loading mechanism.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Abbreviations: CTF, contrast transfer function; ds, double-stranded; PCNA, proliferating cell nuclear antigen; Pfu, Pyrococcus furiosus; priDNA, primed template DNA; RFC, replication factor C; RFCL, RFC large subunit; RFCS, RFC small subunit; ss, single-stranded.
T.M. and H.S. contributed equally to this work. 
To whom correspondence should be addressed. E-mail: morikawa{at}beri.or.jp.
© 2005 by The National Academy of Sciences of the USA
|
References |
|---|
|
TopAbstractMethodsResults and DiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg What's this?
Related Commentary in PNAS:
This article has been cited by other articles in HighWire Press-hosted journals:
|
|
 |
|
  S. Chen, M. M. Coman, M. Sakato, M. O'Donnell, and M. M. Hingorani Conserved residues in the {delta} subunit help the E. coli clamp loader, {gamma} complex, target primer-template DNA for clamp assembly Nucleic Acids Res., April 19, 2008; (2008) gkn157v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Tori, M. Kimizu, S. Ishino, and Y. Ishino DNA Polymerases BI and D from the Hyperthermophilic Archaeon Pyrococcus furiosus Both Bind to Proliferating Cell Nuclear Antigen with Their C-Terminal PIP-Box Motifs J. Bacteriol., August 1, 2007; 189(15): 5652 - 5657. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. G. Anderson, C. R. Williams, M. O'Donnell, and L. B. Bloom A Function for the {psi} Subunit in Loading the Escherichia coli DNA Polymerase Sliding Clamp J. Biol. Chem., March 9, 2007; 282(10): 7035 - 7045. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. R. Barry and S. D. Bell DNA Replication in the Archaea Microbiol. Mol. Biol. Rev., December 1, 2006; 70(4): 876 - 887. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Johnson, N. Y. Yao, G. D. Bowman, J. Kuriyan, and M. O'Donnell The Replication Factor C Clamp Loader Requires Arginine Finger Sensors to Drive DNA Binding and Proliferating Cell Nuclear Antigen Loading J. Biol. Chem., November 17, 2006; 281(46): 35531 - 35543. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Y. Yao, A. Johnson, G. D. Bowman, J. Kuriyan, and M. O'Donnell Mechanism of Proliferating Cell Nuclear Antigen Clamp Opening by Replication Factor C J. Biol. Chem., June 23, 2006; 281(25): 17528 - 17539. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. G. Duggin and S. D. Bell The Chromosome Replication Machinery of the Archaeon Sulfolobus solfataricus J. Biol. Chem., June 2, 2006; 281(22): 15029 - 15032. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. O'Donnell Replisome Architecture and Dynamics in Escherichia coli J. Biol. Chem., April 21, 2006; 281(16): 10653 - 10656. [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Zhuang, B. L. Yoder, P. M. J. Burgers, and S. J. Benkovic The structure of a ring-opened proliferating cell nuclear antigen-replication factor C complex revealed by fluorescence energy transfer PNAS, February 21, 2006; 103(8): 2546 - 2551. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Jeruzalmi The opened processivity clamp slides into view PNAS, October 18, 2005; 102(42): 14939 - 14940. [Full Text] [PDF]
|
|
|
|
|
|
S. L. Kazmirski, Y. Zhao, G. D. Bowman, M. O'Donnell, and J. Kuriyan Out-of-plane motions in open sliding clamps: Molecular dynamics simulations of eukaryotic and archaeal proliferating cell nuclear antigen PNAS, September 27, 2005; 102(39): 13801 - 13806. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
