DNA Polymerase Fidelity: Beyond Right and Wrong

Summary

Accurate DNA replication depends on the ability of DNA polymerases to discriminate between correctly and incorrectly paired nucleotides. In this issue of Structure, Batra and co-workers show the structural basis for why DNA polymerases do not efficiently add correctly paired nucleotides immediately after incorporating incorrectly paired ones.

Genome stability depends upon highly accurate DNA replication. In eukaryotes, the DNA replication machinery makes errors at rates as low as one in 10¹⁰ nucleotides of DNA synthesized. This remarkable accuracy is attributable to three factors (Ganai and Johansson, 2016). The first is nucleotide selectivity – the ability of DNA polymerases to discriminate between correct and incorrect nucleotides during DNA synthesis. The second is proofreading – the ability of polymerase-associated exonucleases to remove incorrect nucleotides from the DNA immediately after their incorporation. The third is mismatch repair – the process by which misincorporated nucleotides are excised from the newly synthesized DNA shortly after replication. Here we will focus on the structural and mechanistic features of DNA polymerases that allow them the synthesize DNA with high fidelity.

Mammalian DNA polymerase beta (pol β), which is involved in base excision repair, is the smallest eukaryotic polymerase. This makes it an ideal system to address fundamental questions about DNA polymerase fidelity. Over the last two decades or so, extensive structural and kinetic studies have led to an ever sharpening image of how this enzyme (and DNA polymerases in general) catalyzes nucleotide incorporation (Beard and Wilson, 2006). The polymerase initiates cycles of processive nucleotide incorporation by binding to the DNA primer-template in an open conformation (Fig. 1). Next, the incoming deoxynucleoside triphosphate (dNTP) binds, and the enzyme undergoes a conformational change to a closed state. In the case of pol β, this change entails movement of the N subdomain so that the nascent base pair (i.e., the base pair formed between the incoming dNTP and the templating nucleotide) is sandwiched between α-helix N and the primer-terminal base pair. Formation of the phosphodiester bond then occurs using a metal-assisted mechanism. A metal ion (usually magnesium) facilitates the deprotonation of the 3′ hydroxyl group of the primer terminus, and the resulting oxyanion carries out a nucleophilic attack on the α-phosphate of the incoming dNTP. The polymerase then opens, advances one nucleotide along the template, and repeats the cycle.

Figure 1.

Comparison of the structures of the binary complex of pol β and DNA in the open conformation (left) and the ternary complex of pol β, DNA, and a correct dNTP in the closed conformation (right).

One of the key questions regarding the mechanism of DNA polymerases is how they discriminate between correct and incorrect incoming dNTPs. This is particularly challenging, because the polymerase must strongly favor only one of the four dNTPs while being able to change its preference in the next cycle when the identity of the templating nucleotide changes. The prevailing view regarding how DNA polymerases selectively incorporate the correct incoming dNTP is the induced fit model (Johnson, 2008). According to this view, conformational changes in the polymerase place the 3′ hydroxyl group of the primer terminus, the α-phosphate of the incoming dNTP, and the catalytic metal ion in the ideal orientation for phosphodiester bond formation only when the incoming dNTP is properly base paired with the templating nucleotide. When the incoming dNTP is mispaired with the templating nucleotide, these atoms are placed in non-optimal arrangements thereby reducing the rate of bond formation.

Studies comparing the structures of ternary complexes comprised of pol β, DNA, and either the correct or the incorrect incoming dNTP have supported the induced fit model (Batra et al., 2006; Batra et al., 2008; Freudenthal et al., 2013; Sawaya et al., 1997). In the presence of both the correct and incorrect dNTPs, pol β is in the closed conformation. In the structure with the correct dNTP, the 3′ hydroxyl group, the α-phosphate, and the metal ion are properly positioned. In the structure with the incorrect dNTP, the conformation of the protein is largely unchanged relative to the structure with the correct dNTP, but the DNA substrate is distorted. The template strand is shifted upstream (away from α-helix N), and the primer terminal residue rotates. This results in a non-optimal orientation of the 3′ hydroxyl group and the α-phosphate and a slow rate of phosphodiester bond formation.

Recently a series of time-resolved structures of pol β have provided snapshots of this enzyme in the act of incorporating correct and incorrect nucleotides (Freudenthal et al., 2013). One of the striking findings to emerge from these structures is a picture of what happens immediately after an incorrect nucleotide has been incorporated. After a correct nucleotide has been added, only minor structural changes are observed. By contrast, after an incorrect nucleotide has been incorporated, the structure changes significantly. The enzyme adopts a somewhat more open conformation and the newly incorporated incorrect nucleotide rotates away from the active site. These changes likely occur because of a buildup of strain in the DNA following misincorporation.

In this issue of Structure, Batra and co-workers (literally) took these studies one step further. They examined what happens structurally when the polymerase attempts to add the next correct dNTP following a misincorporation event. It has been known that DNA polymerases do not efficiently extend from mismatched primer-termini (Beard et al., 2004). In fact, this inefficient extension from a mismatch is what promotes proofreading. To examine the structural basis for how pol β disfavors extending from primer-terminal mismatches, the authors determined the structures of binary complexes comprised pol β and DNA substrates containing 11 of the 12 possible primer-terminal mispairs. (The structure of the binary complex with DNA containing a C-C mispair was not determined due to poor electron density.) They also determined the structures of pol β in ternary complexes with DNA substrates containing all 12 possible primer-terminal mispairs and the next correct incoming dNTP.

In all 11 structures of pol β in binary complexes with DNA substrates containing mismatched primer termini, the enzyme is in the open conformation. The strain resulting from the mismatch at the primer terminus is accommodated by movement of the template backbone at this position away from where it is observed in the binary complex with matched DNA substrates. When the next correct dNTP binds, the enzyme closes. This forces the template residue of the primer-terminal base pair to move into a similar position as seen with the matched substrate. The strain is accommodated in ways that differ for each specific mismatch, but usually results in non-optimal orientations of the 3′ hydroxyl group and the α-phosphate. One of the most extreme examples is when guanine is at the primer terminus and is mispaired with adenine. In the open, binary complex, these bases are both in the usual anti configuration forming two non-Watson-Crick hydrogen bonds. In the ternary complex, when the incoming dNTP binds and the enzyme closes, the primer terminal guanine flips into the abnormal syn conformation. This rotates the sugar and backbone at this position and moves the 3′ hydroxyl group more than 8 Å away from the α-phosphate.

These structures further support the induced fit model. Moreover, they show how this model not only accounts for the inefficient incorporation of nucleotides when a mismatch is present in the nascent base pair, but also accounts for inefficient incorporation when a mismatch is present in the primer-terminal base pair. What is striking about these structures (and the many earlier structures of pol β) is that the precise details of the structural perturbations in the closed, ternary complexes that disfavor incorporation in the presence of mismatches is quite varied. The position of the mismatch, the identity of the mismatched nucleotides, and the broader sequence context all lead to quite different perturbations. Yet this diverse set of perturbations has the same ultimate consequence: failure to properly position the 3′ hydroxyl group, the α-phosphate, and the catalytic metal for efficient catalysis. The emergence of an active site that can employ such varied strategies to disfavor unwanted nucleotide incorporation is truly a marvel of molecular evolution.