Structure of the PCNA unloader Elg1-RFC

Abstract

During DNA replication, the proliferating cell nuclear antigen (PCNA) clamps are loaded onto primed sites for each Okazaki fragment synthesis by the AAA⁺ heteropentamer replication factor C (RFC). PCNA encircling duplex DNA is quite stable and is removed from DNA by the dedicated clamp unloader Elg1-RFC. Here, we show the cryo-EM structure of Elg1-RFC in various states with PCNA. The structures reveal essential features of Elg1-RFC that explain how it is dedicated to PCNA unloading. Specifically, Elg1 contains two external loops that block opening of the Elg1-RFC complex for DNA binding, and an "Elg1 plug" domain that fills the central DNA binding chamber, thereby reinforcing the exclusive PCNA unloading activity of Elg1-RFC. Elg1-RFC was capable of unloading PCNA using non-hydrolyzable AMP-PNP. Both RFC and Elg1-RFC could remove PCNA from covalently closed circular DNA, indicating that PCNA unloading occurs by a mechanism that is distinct from PCNA loading. Implications for the PCNA unloading mechanism are discussed.

Fig. 1.. PCNA clamp unloading by the Elg1-RFC unloader.

(A) Comparison of PCNA loading by RFC and PCNA unloading by Elg1-RFC. How Elg1-RFC unloads PCNA from dsDNA (i.e., after DNA replication) remains unknown. (B) Domain architecture of Elg1-RFC, Rfc2-5, and the PCNA clamp. The Rfc2-5 subunits each contain a AAA⁺ module and a C-terminal “collar” domain. Elg1 also contains these elements but has an insertion in the collar domain that forms the plug inside the central chamber and has two large locking loops, LL1 and LL2 (purple), that prevent conformation changes necessary for PCNA opening and DNA binding. Each PCNA monomer has two structurally similar globular domains [N-terminal domain (NTD) and C-terminal domain (CTD)] that give the PCNA trimer ring a sixfold pseudo-symmetry; they are linked by the inter-domain connecting loop (IDCL). (C) Elg1-RFC is competent for PCNA unloading but not loading. RFC is adopted as a positive control for the ³²P-PCNA loading assay, and a plasmid DNA pUC19 was used to detect the loading and unloading activities of Elg1-RFC for the ³²P-labeled PCNA clamp. Reactions were gel-filtered to separate the large ³²P-PCNA-DNA complex from the smaller “free” ³²P-PCNA (see Materials and Methods for details). Three independent experiments were performed for both the top and bottom experiments, and data points are presented as mean (filled circles) ± SD (error bars). (D) The 3.2-Å resolution EM map and atomic model of Elg1-RFC colored by subunits. The position of the A-gate (between the Elg1 AAA⁺ module and A′ domain) is also labeled. The locking loop density is invisible at the high surface rendering threshold used here. (E) The atomic models of Elg1-RFC complexed with a cracked or a closed PCNA ring. For clarity, the Elg1-RFC pentamer is colored ivory.

Fig. 2.. The A-gate in Elg1-RFC is locked in the closed state by two Elg1 loops.

(A) The front (left) and top (top right, 2.5× enlarged) view of Elg1-RFC (PCNA unloader). Compared with Rfc1 subunit, the Elg1 has an additional loop LL1 to enhance the locking of the A-gate in the closed state. The Elg1 LL2 is equivalent to the Rfc1 alternative linker that can partially unfold to allow the opening of the A-gate (8, 9, 79). The local density of five bound nucleotides and LL1 and LL2 is shown as meshes, and the Elg1 plug is boxed by a blue dashed line and enlarged in middle left. The PCNA ring is closed and is omitted in the top view for clarity. (B) A diagram showing that the A-gate in the RFC clamp loader can be opened to bind both shoulder and central chamber DNA to accommodate the loading of PCNA clamp, while infeasible in the PCNA unloader, Elg1-RFC, due to the Elg1 locking loops and Elg1 plug.

Fig. 3.. The stabilization of the two Elg1 locking loops.

(A) Left: The two linkers of Elg1 are shown as LL1_N and LL1_C (magenta) and LL2 (orange), compared with the alternative linker of Rfc1 (right, orange) that corresponds to the LL2 linker of Elg1. (B to D) The stabilization of Elg1 LL1_N (B), LL1_C (C), and LL2 (D). (B) Elg1 LL1_N (in cartoons) wraps around the collar (in semitransparent surface view, also see Fig. 2A) and is mainly stabilized by two hydrophobic pockets that are demarcated by blue dashed circles and overlaid as hydrophobicity surface plots. (C and D) The LL1_C (C) and LL2 (D) (i.e., positioned above the A-gate) are mainly stabilized by hydrophilic interactions. Key residues involved are shown as sticks, and H-bonds as cyan dashed lines.

Fig. 4.. The Elg1 collar domain has evolved a plug that occupies the central chamber of the Elg1-RFC unloader.

(A) Comparison of full length Elg1 (top) with Rfc1 (bottom). Elg1 and Rfc1 are colored by domain. (B) The five α helices in the collar domain of Elg1 (top) are compared with the Rfc1 collar domain (bottom); two of these α helices are flipped ~180° downward to form the Elg1 plug that occupies the Elg1-RFC central chamber and helps exclude DNA from the central chamber. (C) Topological comparison of the collar domains of Elg1 (top) versus Rfc1 (bottom) reveals that the Elg1 collar domains have rearranged to form the plug that inserts between the AAA⁺ modules and occludes DNA from the central DNA binding chamber (also see movie S2). (D and E) The two boxed regions of Egl1 in (B) are enlarged and placed in context of the Elg1-RFC pentamer. The AAA⁺ domains of Rfc3, Rfc4, and Rfc5 (but not Rfc2) contact and thus stabilize the Elg1 plug. The H-bonds, including intra-molecular ones, are shown as dashed lines in both panels, and the Elg1 plug forms a short parallel β sheet with the Rfc5 plug [i.e., top left of (E)]. See section "The Elg1 plug" for details.

Fig. 5.. Comparison of the interfaces between PCNA and Elg1-RFC and between PCNA and RFC.

(A and B) The binding interfaces of Elg1-RFC (A) and loader RFC (B) on PCNA clamp. The PCNA trimer is shown as surface. For clarity, only the Elg1/RFC regions involved in PCNA binding are shown as cartoons. The interacting subunits and the buried solvent-accessible surface area are labeled. (C) Structural-based sequence alignment suggests a noncanonical PIP motif in Elg1. The structures used are Saccharomyces cerevisiae (S.c.) Elg1-RFC–closed PCNA (this study), Homo sapiens (H.s.) ATAD5 (AlphaFold prediction AF-Q96QE3-F1), S.c. RFC–closed PCNA-DNA [Protein Data Bank (PDB) ID 7TID], S.c. Rad24-RFC–closed 9-1-1 clamp–DNA (PDB ID 7SGZ), and S.c. Ctf18 (AlphaFold prediction AF-P49956-F1). The PIP motifs are boxed by red dashed box, and the consensus feature is shown at the bottom. (D) An enlarged view at the interface of Elg1 and PCNA-1. The PCNA is shown as a transparency surface colored by the local hydrophobicity (left) and cartoon view in ivory (right). Interacting residues are in sticks and labeled (black for PCNA) and green for Elg1 residues. H-bonds are shown as cyan dashed lines.

Fig. 6.. Characterization of Elg1-RFC unloading activity.

(A) Binding of nucleotide is sufficient for Elg1-RFC to unload PCNA from DNA. ³²P-PCNA is first loaded onto nicked plasmid using RFC and then gel-filtered to remove unbound ³²P-PCNA (scheme). Elg1-RFC is added for 2 min with either buffer Elg1-RFC, Elg1-RFC, and ATP, or Elg1-RFC and AMP-PNP. The reaction is then gel-filtered a second time to quantify PCNA remaining on DNA (i.e., fractions 10 to 15), shown as a bar plot to the right. (B) Both RFC and Elg1-RFC can unload PCNA from a covalently closed duplex plasmid. After the first gel filtration, the DNA is sealed with ligase then treated with either RFC or Elg1-RFC followed by gel filtration (scheme). Quantification of PCNA remaining on DNA (i.e., fractions 10 to 15) is shown as a bar plot to the right. The inset within the bar plot is a native agarose gel showing that ligase sealed nicked DNA within 2 min. (C) Pol δ protects PCNA from being unloaded by Elg1-RFC. ³²P-PCNA was loaded onto multi-primed M13mp18 ssDNA coated with RPA and then gel-filtered. Pol δ when then added (or not) for 2 min, followed by Elg1-RFC and ATP for 2 min. Quantification of PCNA remaining on DNA (i.e., fractions 10 to 15) is shown as a bar plot to the right. Some PCNA spontaneously dissociated during the 4-min reaction, as seen in the buffer control, but Pol δ stabilized PCNA on DNA even in the presence of Elg1-RFC. For these gel filtration experiments, duplicate experiments were performed on different days, the individual data points are shown in the histogram, and the primary data are shown in table S3.

Fig. 7.. Hypothetical model of PCNA unloading by Elg1-RFC.

After maturation of an Okazaki fragment, i.e., nick is sealed by DNA ligase, as previously proposed (53, 54), the ligase will dissociate from PCNA to enable Elg1-RFC access to PCNA and unload it from DNA. We propose the Elg1-RFC initially approaches PCNA encircling the dsDNA at a large angle, with the Elg1 subunit first binding PCNA. The unloader then gradually tilts downward (shown by the black curved arrow) to further engage the PCNA. The binding energy likely cracks open the PCNA ring (see also fig. S5 for more details). Because of the presence of the Elg1 locking loops (in the shoulder) and Elg1 plug (in the central chamber), the approaching Elg1-RFC will push the daughter DNA out of the PCNA ring through the open PCNA gate, leading to PCNA unloading and dissociation of Elg1-RFC–PCNA from DNA. Subsequent ATP hydrolysis by Elg1-RFC will lead to separation of Elg1-RFC from the unloaded PCNA, completing the unloading process. Note that the two steps in the dashed box have not been experimentally visualized, probably due to their highly transient nature.