The proofreading mechanism of the human leading-strand DNA polymerase ε holoenzyme

Abstract

The eukaryotic leading-strand DNA polymerase ε (Polε) is a dual-function enzyme with a proofreading 3'-5' exonuclease (exo) site located 40 Å from the DNA synthesizing pol site. Errors in Polε proofreading can cause various mutations, including C-to-G transversions, the most prevalent mutation in cancers and genetic diseases. Polε interacts with all three subunits of the PCNA ring to assemble a functional holoenzyme. Despite previous studies on proofreading of several Pol's, how Polε-or any Pol complexed with its sliding clamp-proofreads a mismatch generated in situ has been unknown. We show here by cryo-EM that a template/primer DNA substrate with a preexisting mismatch cannot enter the exo site of Polε-PCNA holoenzyme, but a mismatch generated in situ in the pol site yields three bona fide proofreading intermediates of Polε-PCNA holoenzyme. These intermediates reveal how the mismatch is dislodged from the pol site, how the DNA unwinds six base pairs, and how the unpaired primer 3'-end is inserted into the exo site for cleavage. These results unexpectedly demonstrate that PCNA imposes strong steric constraints that extend unwinding and direct the trajectory of mismatched DNA and that this trajectory is dramatically different than for Polε in the absence of PCNA. These findings suggest a physiologically relevant proofreading mechanism for the human Polε holoenzyme.

Keywords: DNA polymerase; DNA polymerase epsilon; DNA proofreading; leading strand; replication fidelity.

Fig. 1.

Cryo-EM structures of Polε–PCNA bound to T/P with a preexisting mismatch. (A) sketch of the Polε holoenzyme proofreading process based on published studies in the absence of sliding clamp. The T/P unwinds by 3 bp and the unwound regions are kept separated during mismatch editing. The pol and exo sites are labeled. (B) Domain architectures of PolE1-4 and PCNA. Dashed lines indicate disordered regions. The structurally resolved regions (POLE1-NTD, i.e., Polε-core, and PCNA are colored by domains. POLE1-CTD and POLE2-4 are mobile relative to POLE1-NTD and invisible in EM maps. (C) The nucleotide sequence of the primer and template DNA. The mismatched primer 3′-end is highlighted in red. The T/P with a preexisting mismatch cannot reach the exo site of Polε holoenzyme and binds Polε in a blocked state. (D) EM maps and atomic models of the Polε–PCNA–T/P complex in two blocked conformations. The maps and models are labeled and colored by domains as in (B). The distinct interactions of the Exo and thumb domains in the two conformations are highlighted by the red box.

Fig. 2.

Cryo-EM structures of Polε–PCNA bound to an in situ generated mismatch. (A) Scheme for capturing Polε holoenzyme in the proofreading states. Nucleotide sequence of the P/T is shown in color. In the presence of dTTP, Polε extends 4T; 3 are matched with template (gray) and the last one is mismatched (red). (B–D), Cryo-EM maps and atomic models of the ternary complex in the mismatch-locking (B), Pol-backtracking (C), and mismatch-editing (D) states. Cryo-EM maps are shown in the Upper panel, atomic models in the Middle, and close-up views of the mismatched primer 3′-end in the three proofreading intermediates in the Lower panel.

Fig. 3.

Conformational changes of the T/P from the mismatch-locking to Pol-backtracking state. (A) Superimposition on Polε-core of the mismatch-locking (color) and Pol-backtracking states (gray). Domain movements are indicated by red arrows. (B) A top view of the superimposition at the Polε–PCNA interface where most movements occur. The bulk of Polε-core is omitted for clarity. (C) Close-up view of changes in the mismatched region of T/P between the two states. The movements are indicated by red arrows and labeled. (D) Superimposition of T/P based on alignment on PCNA in both states. The DNA tilts 20° and moves with the thumb outward by 9 Å. And the DNA moves down by 1 bp.

Fig. 4.

Polε interactions with T/P DNA in the mismatch-editing state. (A) The mismatch-editing state structure colored by domains. (B) T/P DNA in sticks superimposed with the EM density in transparent surface view. The top orange box shows the clear DNA density around the exo⁻ site of Polε, the bottom blue box shows the weak DNA density inside the PCNA clamp. (C) Electrostatic surface of the exo⁻ site with the primer 3′-end. (D) The Polε exo⁻ site structure. The catalytic D275 and E277 coordinating two Mg²⁺ in wild-type Polε are shown; they are mutated to A275 and A277 in Polε-exo⁻. (E) Close-up view of T/P around the exo^- site. The T/P melts 6 bp then forms four out-of-register base pairs. The β-hairpin does not contact the template. Residues stabilizing the T/P are in sticks and labeled. (F) Sketch of T/P in editing state. (G) Comparison of the short Polε β-hairpin of (dark orange) and the longer RB69 gp43 β-hairpin (gray). The longer phage β-hairpin separates the template from primer.

Fig. 5.

Conformational changes of Polε–PCNA induce the mismatched DNA translation during backtracking. (A) Comparison of the Pol-backtracking (gray) and mismatch-editing (color) states aligned on Polε-core. (B) Close-up view of the T/P and surrounding region. Red arrows indicate movements between the two states. Key residues are in sticks and labeled. (C) Structural changes near the Polε–PCNA interface. The top region of Polε-core is omitted for clarity. (D) Side-by-side comparison of the P-domain and T/P interaction in the two states. DNA-interacting residues are in sticks and labeled. (E) Comparison of the T/P in the two states reveals an overall translocation. The DNA tilts 10° toward the exo activity site. The duplex moves upward allowing the unwinding of 6 bp from the primer 3' end, thereby enabling the primer to move 15 Å through exo channel into the exo site.

Fig. 6.

Model of the Polε–PCNA holoenzyme proofreading process. (A) Schematic representation of all five steps that are built based on experimental structures of the human Polε–PCNA holoenzyme. In step 1, Polε incorporates a mismatched base in the pol site and senses the mismatch via a Watson–Crick base-pairing checkpoint. This is modeled by Polε structure in the Pol state (PDB ID 9B8T). In step 2, Fingers domain flips up to an open position to arrest the pol activity, and the holoenzyme moves away from the mismatched 3′-end by 1 bp to prevent additional base incorporation. In step 3, the Polε linker helix rotates 10° and the thumb domain moves outward by 9 Å. The movements put pressure on the 1-bp backtracked T/P. In step 4, the P-domain tilts 12° against the PCNA, and the linker helix rotates 36°, causing the holoenzyme to backtrack by 6 bp and unwind 6 bp from the primer 3′-end. The six unwound bp rebind and form four out-of-register bp and to insert the mismatched primer 3′-end into the exo site. In step 5, the primer 3′ mismatch is excised by the exo activity. Next, we assume the holoenzyme returns the T/P to the pol site and resumes DNA synthesis. (B) Schematic describing the mismatch DNA translocation during the Polε–PCNA holoenzyme proofreading process. The horizontal dashed line indicates the postinsertion site of Polε. The movement of DNA between each step is indicated by red arrows.