New insights into DNA polymerase mechanisms provided by time-lapse crystallography

Abstract

DNA polymerases play central roles in DNA replication and repair by catalyzing template-directed nucleotide incorporation. Recently time-lapse X-ray crystallography, which allows one to observe reaction intermediates, has revealed numerous and unexpected mechanistic features of DNA polymerases. In this article, we will examine recent new discoveries that have come from time-lapse crystallography that are currently transforming our understanding of the structural mechanisms used by DNA polymerases. Among these new discoveries are the binding of a third metal ion within the polymerase active site, the mechanisms of translocation along the DNA, the presence of new fidelity checkpoints, a novel pyrophosphatase activity within the active site, and the mechanisms of pyrophosphorolysis.

Introduction

DNA polymerases play central roles in DNA replication, recombination, and repair. Thus, they play an essential biological role in maintaining genome stability. These enzymes are divided into several families based on their evolutionary relationships (Table 1) [1,2]. A and B family polymerases are involved in DNA replication and repair and are found in both prokaryotes and eukaryotes. C and D family polymerases are involved in DNA replication and are found in prokaryotes and archaea, respectively. X family polymerases are involved in DNA repair and are found in eukaryotes. Y family polymerases are involved in translesion synthesis and are found in both prokaryotes and eukaryotes.

Table 1.

DNA polymerase families

Family	Domains	Examples	Functions
A	Bacteria Eukarya	E. coli pol I H. sapiens pol γ, pol θ	DNA replication and DNA repair
B	Bacteria Archaea Eukarya	E. coli pol II H. volcanii Dpo2, Dpo3 H. sapiens pol α, pol δ, pol ε	DNA replication and DNA repair
C	Bacteria	E. coli Pol III	DNA replication
D	Archaea	H. volcanii Pol D	DNA replication
X	Eukarya	H. sapiens pol β, pol λ, pol μ	DNA repair
Y	Bacteria Archaea Eukarya	E. coli pol IV, Pol V H. volcanii Dpo4 H. sapiens pol η, pol ι, pol κ, REV1	Translesion synthesis

DNA polymerases carry out template-directed nucleotide incorporation using the same basic reaction scheme (Fig. 1) [3] . First, the polymerase binds DNA to form the polymerase-DNA binary substrate complex. This binary complex then binds the incoming dNTP and two divalent metal ions to form the polymerase-DNA-dNTP ternary substrate complex (Fig. 1, left panel). Next, the 3’-OH is deprotonated to activate the primer terminal oxygen and promote nucleophilic attack generating the transition state complex (Fig. 1, middle panel). The resulting chemical step of phosphodiester bond formation results in the incorporation of the incoming nucleotide to form the polymerase-DNA-pyrophosphate ternary product complex (Fig. 1, right panel). The pyrophosphate is then released to form the polymerase-DNA binary product complex. Finally, either the polymerase translocates one nucleotide along the DNA template to reform the polymerase-DNA binary substrate complex or the polymerase disengages from the DNA substrate. Note that this is a minimal reaction scheme as additional steps including conformational changes, metal ion binding/release, and pyrophosphate breakdown, have been shown to occur in specific polymerases (highlighted below).

Figure 1.

Schematic of the conventional DNA polymerase two-metal mechanism for nucleotide incorporation. The ternary, transition state, and product complex are shown with the primer terminus, incoming dNTP, metal ions, and catalytic triad indicated. The orange arrow denotes the nucleophilic attack following deprotonation of the 3’-OH.

Conventional X-ray crystallography has provided enormous insights into the structures and mechanisms of DNA polymerases [2-6]. These enzymes possess a polymerase domain, which is often divided into fingers, thumb, and palm subdomains [7]. Many polymerases also possess accessory domains such as a 3´ to 5´ exonuclease domain for proofreading, a 5´ to 3´ exonuclease domain for nick translation, and a lyase domain for processing DNA repair intermediates [8-10]. In addition, many structures have been determined of DNA polymerases bound to DNA substrates and bound to DNA and incoming dNTP substrates [2]. These have contributed greatly to our understanding of the conformational changes that can occur upon substrate binding, the roles of two metal ions in promoting catalysis, the structural basis of polymerase fidelity, the mechanism of proofreading the incorrectly inserted nucleotide, and the ability of some polymerases to accommodate damaged DNA or dNTP substrates.

Time-lapse X-ray crystallography allows one to observe reaction intermediates that are normally difficult to capture at the molecular level (Fig. 2). This technique has revealed numerous and unexpected mechanistic features of DNA polymerases including the binding of a third metal ion within the active site, the translocation of the DNA polymerase along the DNA substrate, the discovery of general and enzyme-specific fidelity checkpoints during nucleotide incorporation, a novel pyrophosphatase activity occurring within the active site, and a mechanistic description of pyrophosphorolysis (the reverse rection) [11-17]. Here, we examine recent discoveries contributing to our understanding of DNA polymerase mechanisms that have come from the utilization of time-lapse crystallography. These discoveries are currently transforming our understanding of DNA polymerase mechanisms and are raising important new questions that require future investigation. While this article focuses on DNA polymerases, the time-lapse crystallography approach has been utilized with other enzyme systems to observe novel reaction intermediates [18-22].

Figure 2.

Schematic of the time-lapse X-ray crystallography approach. Cartoon representations of the binary, ternary, and product polymerase: DNA complexes are shown in blue. The DNA polymerase is crystallized with a DNA substrate to generate the binary complex. The binary complex crystals are transferred to a cryo-solution containing dNTP substrates and Ca²⁺ ions, denoted by the number 1. The ternary complex crystals are then transferred to a second cryo-solution containing a metal (Mg^2+/Mn²⁺) that initiates catalysis, denoted by the number 2. After initiating catalysis, the crystals are flash frozen at different time points and X-ray diffraction data collected to obtain structural snapshots of the DNA polymerase during and post catalysis.

Time-lapse X-ray crystallography of DNA polymerases

Time-lapse X-ray crystallography is like other time-resolved crystallography approaches in that it allows one to capture key structural intermediates during the polymerase reaction (Fig. 2) [23-26]. This is typically accomplished by crystallizing the DNA polymerase in complex with DNA to generate a binary complex. These binary complex crystals are then soaked in a cryo-solution with natural dNTP substrates with Ca²⁺ ions, which promote the formation of polymerase-DNA-dNTP complexes (i.e., the substrate ternary complexes), but do not promote catalysis. The resulting crystals are then transferred to a cryo-solution contain Mg²⁺ for various lengths of time before being flash frozen in liquid N₂. During this time interval, the Ca²⁺ ions are replaced by Mg²⁺ ions, which promote nucleotide incorporation forming polymerase-DNA-PP_i complexes (i.e., the product ternary complexes). Following flash freezing, X-ray diffraction data is collected to obtain structural snapshots at various time intervals.

Mechanism of nucleotide incorporation

Conventional X-ray crystallography has shown that two metal ions (typically Mg²⁺ ions) are found within the active site of DNA polymerases [2,27,28]. This led to the view that nucleotide incorporation is catalyzed by a two-metal mechanism [29,30]. One metal ion, which binds at the A-site, facilitates deprotonation of the 3' OH of the primer strand allowing the oxygen to act as a nucleophile to attack the α-phosphate of the incoming dNTP (orange arrow, Fig 1). Another metal ion, which binds at the B-site prior to metal A, is thought to stabilize the negative charge that accumulates on the non-bridging oxygen of the α-phosphate during the transition state (middle panel, Fig.1).

Initial time-lapse crystallography work discovered that a third metal ion binds at a novel site near the α and β phosphates of the incoming dNTP (the C-site) in the case of several DNA polymerases (Fig. 3a). For example, a third metal ion is observed with X-family polymerases pol β, pol λ, and pol μ, and with Y-family polymerase pol η, but was not observed with Y-family polymerase Rev1[12,14,31,32]. Furthermore, the third metal ion was observed during both correct and incorrect nucleotide incorporation by pol η [16]. By contrast, a third metal ion was only observed during correct nucleotide incorporation by pol β and pol λ, but not during incorrect nucleotide incorporation [31,33]. This use of a third metal ion during catalysis has been termed the three-metal mechanism and has been recently observed during time-lapse crystallographic studies of other metal-dependent enzymes[18,34].

Figure 3.

Examples of structural insights into DNA polymerase mechanisms obtained by time-lapse X-ray crystallography. (a) DNA pol β product complex structure highlighting the product-associated third metal Me_C (left, PDB:4KLG). DNA pol η intermediate complex structure highlighting the transition-state third metal Me_C (right, PDB:4ECV). (b) Bst pol binary complex structure in the pre-translocation conformation prior to incoming nucleotide binding and insertion (left, PDB:7K5P). Bst pol ternary complex structure following translocation and binding the dNTP (right, PDB:7K5Q, metal ions are from PDB:1LV5). (c) DNA pol η ternary complex structure with a correct incoming nucleotide (left, PDB:4ECQ). DNA pol η ternary complex structure with an incorrect incoming nucleotide highlighting conformational changes in the 3-OH that reduces nucleotide misincorporation efficiency (right, PDB:7U72). (d) DNA pol μ intermediate complex structure during 8-oxo-dGTP insertion across from templating C (left, PDB:7KTB). DNA pol μ intermediate complex structure during 8-oxo-dGTP insertion across from templating A (right, PDB:7KT4). (e) DNA pol IV product complex structure with pyrophosphate in the active site (left, PDB:5YV0). DNA pol IV product complex structure with two monophosphates in the active site following pyrophosphate hydrolysis (right, PDB: 5YV2). (f) DNA pol β ternary complex structure with PNP in the active site (left, PDB:5UGO). DNA pol β product complex structure with dNMPP_NP in the active site following pyrophosphorolysis (right, PDB:5UGN). For all structures, the DNA is shown as grey or yellow sticks and active site metals are shown as green spheres. Key active site changes are highlighted.

The electron density for the third metal ion accumulates with the similar kinetics as the chemical step of phosphodiester bond formation. This makes it difficult to determine whether the third metal binds immediately before or immediately after phosphodiester bond formation. This has led to significant controversy about the function of the third metal ion during nucleotide incorporation [35-37]. Studies with pol η have led to the proposal that the third metal ion binds immediately prior to the nucleotide incorporation and promotes catalysis by driving bond breaking between the α and β phosphates [14,15]. Studies with pol β , pol λ, and pol μ have led to the proposal that the third metal ion binds immediately after nucleotide incorporation and is associated with the amount of product formed [12,31,33]. In this case, it has been hypothesized to serve as a metal co-factor that deters pyrophosphorolysis (the reverse reaction) [38]. While time-lapse crystallography resulted in the important discovery of a third metal ion in the DNA polymerase active site, there remains many unanswered questions regarding the precise function of this metal ion during catalysis. Future integrated structural, computational, and biochemical studies are needed to elucidate the exact role of the third metal ion during DNA polymerase catalysis.

Mechanism of translocation

The chemical step of phosphodiester bond formation is widely believed to occur prior to the translocation step. In fact, time-lapse crystallography with pol β, pol η, pol μ, pol λ, Bacillus pol I, and Rev1 provide strong evidence that the chemical step does precede the translocation step [12,14,31-33,39,40]. However, a recent study with G. stearothermophilus BST, a bacterial A family replicative polymerase, proposed that this sequence of events may not be universal. A series of structural snapshots showed that the template base in the polymerase-DNA binary complex is initially outside the active site. The templating base flips into the active site guided by the finger subdomain as the reaction proceeds, which enables binding of the incoming dNTP (Fig 3b). This conformational change corresponds to a translocation of the polymerase along the DNA substrate prior to phosphodiester bond formation. This suggests that in the case of some DNA polymerases, the translocation step may precede the chemical step of nucleotide incorporation.

This conclusion, however, remains controversial, as it is difficult to understand how catalysis can occur given the positions of the reactive groups and the metal ions in the structure of translocated substrate ternary complex (Fig. 3b) compared to a more typical pre-translocation substrate ternary complex of this same enzyme determined by a different group [40]. Thus, it remains possible that the translocated substrate ternary complex represents an intermediate that forms but is not catalytically competent. While time-lapse crystallography allowed the visualization of this novel intermediate state, this highlights how it is often difficult to interpret how these intermediate states contribute to catalysis. Ultimately, more work is needed to understand whether this polymerase genuinely translocates prior to nucleotide incorporation and if so, to determine how widespread such a mechanism is.

Structural basis of fidelity

DNA polymerase fidelity is the specificity for incorporating the incoming dNTP that forms correct Watson-Crick base pairs with the template base versus the other incorrect dNTPs that do not form proper Watson-Crick base pairing [41,42]. This fidelity is achieved by two means. First is the intrinsic specificity of the polymerase activity for incorporating the correct nucleotide rather than the incorrect nucleotides [43]. Second is the ability of many DNA polymerases to proofread their products using a 3' to 5' exonuclease activity. Together these two factors allow polymerases to achieve error frequencies as low a 1 in 10⁶ to 10⁸ nucleotides [44].

Conventional X-ray crystallography has contributed greatly to our understanding of DNA polymerase substrate specificity that leads to incorporating nucleotides with high fidelity [43,45]. When bound to a correct dNTP, the enzyme generally favors nucleotide incorporation by aligning active site residues, the 3' OH of the primer strand of the DNA substrate, and the α-phosphate of the incoming dNTP substrate leading to efficient catalysis. Likewise, when bound to an incorrect dNTP, the enzyme disfavors nucleotide incorporation by misaligning active site residues and the DNA substrates leading to inefficient catalysis.

Initially, the time-lapse crystallography approach was used with pol β and pol η to obtain insight into polymerase fidelity and the checkpoints that ensure the correct nucleotide is incorporated [14,31]. More recent time-lapse crystallography studies have also led to additional mechanistic insights into DNA polymerase fidelity. For example, in the case of Y-family polymerases pol η, binding of the incorrect nucleotide resulted in improper base pairing between the incoming nucleotide and template residue [16]. This results in suboptimal alignment of the 3' OH of the primer strand involved in phosphodiester bond formation (Fig. 3c). This suboptimal alignment reduces the efficiency of misincorporation of the incorrect nucleotide and serves as a key checkpoint for fidelity, as this has also been observed for an additional Y-family polymerase Rev1 [45]. Additional time-lapse crystallography with X-family polymerase pol λ led to the identification of new fidelity checkpoints [33]. These include conformational changes in the DNA template, active site residues, and loops that resulted in a deformed active site geometry that reduces the efficiency of nucleotide insertion.

Accommodation of DNA damage

Conventional X-ray crystallography has provided many insights into the mechanisms used by DNA polymerases to accommodate DNA damage during nucleotide incorporation [2]. For example, several studies have identified that replicative and repair polymerases efficiently incorporate either dCTP or dATP opposite 8-oxo-dG, a common type of oxidative base damage that is often utilized efficiently by many polymerases [46]. Additional studies have also determined how DNA polymerases accommodate DNA damage in the templating base position during lesion bypass [2]. However, these studies have been unable to address the role of a third metal ion and/or active site rearrangements that occur during incorporation or bypass of DNA damage to allow DNA polymerases to avoid fidelity checkpoints.

Time-lapse X-ray crystallography has provided snapshots of DNA polymerases during the incorporation of the oxidized incoming nucleotide 8-oxo-dGTP opposite either template A or template C by X-family polymerase pol β [11]. 8-oxodGTP is incorporated in the syn configuration opposite A, which is stabilized by specific contacts with the DNA polymerase active site. In contrast, 8-oxo-dGTP is incorporated in the anti-configuration opposite C. This configuration is stabilized by the appearance of a third metal ion, which alleviates a clash between O8 and α-phosphate of the incoming dNTP. Recent time-lapse crystallography with the X-family DNA polymerase pol μ provided snapshots that rationalize how 8-oxo-dGTP is efficiently incorporated opposite template C, but not template A [47]. During incorporation of 8-oxo-dGTP (in the anti-configuration) opposite C, a third metal ion was also observed (Fig. 3d). Importantly, this third metal ion is not observed for incorporation of 8-oxo-dGTP (in the syn configuration) opposite A. Time-lapse crystallography was also recently utilized to investigate insertion of the ribo-8-oxo-GTP by pol μ to determine the mechanisms by which this damaged nucleotide avoids polymerase discrimination checkpoints [47].

Mechanisms of polymerase-associated reactions

Time-lapse crystallography was recently used to examine the mechanism of nucleotide incorporation by Y-family pol IV from Escherichia coli and the Y-family polymerase Rev1 from Saccharomyces cerevisiae [32,48] . Both studies identified structural snapshots of two P_i products in the active site of the product complex, instead of a PP_i product (Fig. 3e). This indicates that bond cleavage had occurred between the β- and γ-phosphates in the polymerase active site, consistent with an intrinsic pyrophosphatase activity. This pyrophosphatase activity is biologically significant for two reasons. First, the cleavage of the PP_i to two P_i molecules shifts the equilibrium of the nucleotide incorporation reaction in the forward direction and helps drive this reaction thermodynamically [49]. Second, the cleavage PP_i also favors nucleotide incorporation by disfavoring pyrophosphorolysis (the reverse reaction) [50]. Interestingly, this pyrophosphatase activity was not observed in time-lapse crystallographic studies with pol β, pol η, pol λ, and pol μ [12,14,31,33], suggesting that this novel activity may be unique to a subset of DNA polymerases. Several mechanistic questions remain that require further investigation including which active site residues act as a general base for the pyrophosphatase activity and the temporal events of pyrophosphate breakdown in relation to nucleotide incorporation.

Pyrophosphorolysis is the reverse reaction of nucleotide incorporation, which involves the binding of PP_i and the subsequent attack on the phosphate group of the primer-terminal residue to generate dNTP in the active site [51]. This reaction may serve as an intrinsic proofreading mechanism for DNA polymerases that lack a 3´ to 5´ exonuclease domain (repair-associated polymerases) [52]. Time-lapse crystallography was recently used to dissect the mechanism of pyrophosphorolysis by X-family polymerase pol β [53]. These studies used a PP_i analog (PNP) that replaces the non-bridging oxygen atom of the PP_i with a nitrogen atom, which changes the active site equilibrium towards the pyrophosphorolysis reaction. Binding of PNP facilitated binding of two metal ions that are accommodated in the A-site and B-site (Fig. 3f). Interestingly, a third metal was not observed at the C-site during the snapshots obtained for the reverse reaction, consistent with the prior hypothesis that this metal co-factor likely prevents pyrophosphorolysis [12,31]. Future work is needed to fully dissect the role of the third metal and mechanism of pyrophosphorolysis by polymerases.

Conclusions

The mechanisms of nucleotide incorporation by DNA polymerases have been studied for decades. These mechanistic insights have largely come from conventional X-ray crystallography and extensive biochemical analysis. However, the recent application of time-lapse crystallography to DNA polymerases has brought about a new age in the DNA polymerase field that has led to a host of novel discoveries and exciting unanswered/unresolved questions. These include the role of the third metal ion in catalysis, the mechanism(s) of translocation, the identity of new fidelity checkpoints, the mechanisms used to accommodate damaged DNA and dNTP substrates, the mechanism of pyrophosphatase activity, and the mechanisms of pyrophosphorolysis. These novel discoveries and unanswered/unresolved questions require future work to obtain a comprehensive and integrated understanding of the structure and function of DNA polymerases, which will ultimately be facilitated by time-lapse X-ray crystallography.