Structure of DNA-CMG-Pol epsilon elucidates the roles of the non-catalytic polymerase modules in the eukaryotic replisome

Abstract

Eukaryotic origin firing depends on assembly of the Cdc45-MCM-GINS (CMG) helicase. A key step is the recruitment of GINS that requires the leading-strand polymerase Pol epsilon, composed of Pol2, Dpb2, Dpb3, Dpb4. While a truncation of the catalytic N-terminal Pol2 supports cell division, Dpb2 and C-terminal Pol2 (C-Pol2) are essential for viability. Dpb2 and C-Pol2 are non-catalytic modules, shown or predicted to be related to an exonuclease and DNA polymerase, respectively. Here, we present the cryo-EM structure of the isolated C-Pol2/Dpb2 heterodimer, revealing that C-Pol2 contains a DNA polymerase fold. We also present the structure of CMG/C-Pol2/Dpb2 on a DNA fork, and find that polymerase binding changes both the helicase structure and fork-junction engagement. Inter-subunit contacts that keep the helicase-polymerase complex together explain several cellular phenotypes. At least some of these contacts are preserved during Pol epsilon-dependent CMG assembly on path to origin firing, as observed with DNA replication reconstituted in vitro.

Fig. 1

Cryo-EM structure of the tetrameric Pol epsilon complex, lacking the catalytic domain (deltacat). a Subunit composition and domain organization of yeast Pol epsilon. N-Pol2 stands for N-terminal Pol2. C-Pol2 stands for C-terminal Pol2. ZnF stands for zinc-finger appendix. b 2D class averages of deltacat. c Surface rendering of the deltacat structure solved to 4.45 Å resolution. Dpb2 is green, Pol2 catalytically dead polymerase fold is blue, and Pol2 zinc-finger appendix is orange. d Atomic model for the yeast Dpb2 (green) bound to the Pol2 C-terminal zinc-finger appendix (orange), built into the cryo-EM map. e Detail of the Pol2 zinc-finger appendix (further map sharpening with phenix.auto_sharpen)

Fig. 2

Interaction between C-Pol2 and Dpb2. a Atomic model of C-Pol2 and Dpb2 built into the cryo-EM map. b Dpb2 C-terminus is poised towards the C-Pol2 interface, explaining why a truncation of the Dpb2 C-terminal region results in a lethal phenotype. c Pol2 zinc-finger appendix resides in the core of the Pol2 polymerase fold. This explains why a point mutation disrupting ZF1 is not compatible with viability. A second zinc-finger motif (ZF2) projects from the core of the complex

Fig. 3

The C-terminal half of Pol2 contains a catalytically dead polymerase fold that has lost its DNA-binding function. a C-Pol2 contains an inactive polymerase fold with jaws wide open. b Coulombic surface coloring of the C-Pol2 catalytically dead polymerase reveals a lack of positive charges in the vestiges of the DNA-binding groove, while the DNA-binding site in the related Pol alpha catalytic domain is positively charged. c The DNA-binding groove in C-Pol2 is occupied by the Pol2 zinc-finger appendix. d DNA-binding assay with various Pol epsilon variants. Only constructs containing the N-Pol2 catalytic domain or the histone-related proteins Dpb3-Dpb4 bind to DNA. We were unable to detect any DNA binding for C-Pol2 and Dpb2. On the right, quantification of the DNA-binding assays. Each experiment was repeated three times and error bars indicates standard deviation. Also refer to Supplementary Fig. 3

Fig. 4

Cryo-EM structure of CMG-Pol epsilon. a Diagram of the reconstitution of CMG-Pol epsilon on a pre-formed fork. CMG stands for Cdc45-MCM-GINS. b Silver-stained SDS-PAGE gel of the reconstituted CMG-Pol epsilon-DNA complex. c Two-dimensional class averages of the CMG-Pol epsilon-DNA complex. d Cryo-EM structure of CMG-Pol epsilon with docked/real-space refined homology models of CMG and C-Pol2/Dpb2 components

Fig. 5

ATPase state and DNA binding in the CMG-Pol epsilon complex. a Inspection of the local resolution map indicates that the ATPase site clamped by C-Pol2/Dpb2 is highly stable, with the Mcm5 AAA+ module reaching resolutions as high as 4.5 Å. b Single-stranded DNA is captured by ATPase pore loops of Mcm3, -5, -2, and -6. ATPγS is bound to the Mcm5-3 and Mcm2-5 subunits. c A cut-through view of the CMG-Pol epsilon reveals duplex DNA entering through the N-terminal domain of MCM and the leading-strand template captured within the ATPase domain. A 90° rotated view of the Mcm3-5-2 region reveals how C-Pol2/Dpb2 clamping stabilizes nucleotide binding and single-stranded DNA engagement by Mcm2, Mcm5 and Mcm3 pore loops

Fig. 6

Interactions between Pol epsilon and the CMG. a A model for full-length Dpb2. N-terminal and C-terminal domains are tethered by a flexible linker. N-terminal Dpb2 contacts the C-terminal domain of GINS subunit Psf1. b The winged helix domain of Mcm5 is clamped between the catalytically dead polymerase in Pol2 and the zinc-finger appendix. c C-Pol2 and Dpb2 keep the ATPase domains of Mcm2-5-3 in a compacted state. C-Pol2 interacts with Mcm2 (via the polymerase fold) and Mcm5 (via the ZF2 element in the zinc-finger appendix, also see Fig. 2c). Dpb2 contacts the Mcm3 ATPase domain. d A complex containing both C-Pol2 and Dpb2 represents the minimal complement of Pol epsilon modules required for assembly of the CMG. e A complex of C-Pol2/Dpb2 but not the two isolated protomers support DNA replication in vitro

Fig. 7

Composition of the replicative helicase productively engaged to the replication fork. a Diagram of the EM-based translocation assay. Multiple MCM double hexamers are loaded onto a linear stretch of duplex DNA capped with protein–DNA roadblocks. Only certain MCM double hexamers are activated and translocating CMGs push MCM double hexamers against the DNA roadblock. b Representative negative-stain micrograph reveals accumulation of MCM trains (marked with a khaki line) alongside isolated CMGs (marked with purple circles). Scale bar 50 nm. c Detail of a negative-stain micrograph revealing that MCM trains are made of stacked double hexamers loaded onto DNA. d Two-dimensional averages of the CMG capping the trains and opposed end of trains capped by a protein roadblock. Most MCM-pushing CMG particles are bound to Pol epsilon. e 2D averages of CMG-Pol epsilon and MCM double hexamers, next to a cartoon of an MCM blocked by a protein–DNA roadblock. f Atomic structures of CMG-Pol epsilon and MCM double hexamer, both bound to DNA. g 2D class averages of isolated CMGs observed in the MCM train experiment. Most CMGs that failed to translocate up to the roadblock are not engaged by Pol epsilon. h Cryo-electron tomogram of an MCM train. Scale bar 50 nm. i MCM and CMG structures placed into the cryo-electron tomogram using template matching approaches. This experiment supports the notion that CMG-Pol epsilon particles push MCM double hexamers against a protein–DNA roadblock to form MCM trains

Fig. 8

Proposed mechanism for the concomitant recruitment of GINS and Pol epsilon onto MCM. By interacting with Cdc45 on the N-terminal MCM face, GINS stabilizes the Mcm2-5-3 N-terminal interactions. C-Pol2 and Dpb2 in the Pol epsilon complex play a similar role, by stabilizing the Mcm2-5-3 AAA+ ATPase interactions. Given that GINS and Pol epsilon are recruited onto MCM as part of the same complex, concomitant biding of both factors might promote activating conformational changes in the helicase, eventually leading to origin activation