Recognizing that in crystallo and solution catalyzed enzymatic reactions can produce different structural results with potentially different functional interpretations (Ehrmann et al., 2017), we chose to investigate the translocated intermediates of DNAP-I using a direct crystallization method that involves solving crystal structures of the enzyme-product complex obtained from primer-extension reactions performed in solution rather than inside the environment of a protein crystal. In these reactions, the starting enzyme-primer-template complex was incubated with solutions of either buffer, dTTP, or dTTP and dATP for 30 min at 37°C. Following primer-extension, the enzyme-product complex was crystallized and cocrystal structures of Bst DNAP-I were solved to resolutions of 1.5 - 2.0 Å (Table 1). This approach was used to obtain high resolution structures of DNAP-I for the starting primer-template complex (n) and two translocated products obtained for the n + 1 and n + 2 nucleotide addition steps using the same primer-template duplex (n) described in previous in crystallo studies (Figure 1a)(Johnson et al., 2003).
Structures of the enzyme-primer-template complex (n) before catalysis reflect the initiation step of DNA synthesis. Superposition of the new structure obtained for the initiation step against the previously solved structure reveals that both structures adopt the same active site conformation (Figure 1—figure supplement 2a). This result implies that any structural differences observed between the translocated product of solution and in crystallo catalyzed reactions should be due to the catalysis environment rather than the starting polymerase conformation.
To evaluate the elongation step of DNA synthesis, the translocated products obtained from solution and in crystallo catalyzed primer-extension reactions were compared, both globally and locally within the enzyme active site (Johnson et al., 2003; Kiefer et al., 1998). All of the structures adopt the same overall topology commonly observed for A-family DNA polymerases (Figure 1b). However, careful analysis of the enzyme active site did reveal clear conformational differences between structures obtained from solution-catalyzed reactions versus those obtained from in crystallo catalyzed reactions (Figure 1c,d). The in crystallo catalyzed reactions adopt an active site conformation that is nearly identical to the starting conformation, which represents the initiation step of DNA synthesis (Figure 1—figure supplement 2a). However, the solution catalyzed reactions produce a different active site conformation that binds the duplex in a different position and base pair geometry (Figure 1—figure supplement 2b,c).
Major structural differences are depicted in the 2D interaction maps, which show that the solution catalyzed reactions produce a translocated product with markedly fewer contacts to the phosphodiester linkage, sugar, and nucleobase moieties of the primer-template duplex as compared to the translocated product obtained by in crystallo catalysis (Figure 1—figure supplements 3 and 4, Supplementary file 1a). A particularly striking example of conformational disparity is Tyr714, a critical active site residue involved in the mechanism of DNA synthesis (Bell et al., 1997; Carroll et al., 1991). In the solution catalyzed structures, Tyr714 stabilizes the newly formed base pair by stacking above the primer strand, while this residue stacks above the template strand in the in crystallo catalyzed structures (Figure 1c,d). Importantly, the pre-insertion site is not observed in the solution catalyzed reactions due to a kink in the O-helix, which abrogates the O-O1 loop in the finger subdomain (Figure 1d). Absent a hydrophobic pocket, the n + 1 nucleotide in the template strand stacks against Tyr719 in the O1 helix, which positions the base for a subsequent round of catalysis. The solution catalyzed structures obtained for the n + 1 and n + 2 translocated products adopt identical active site conformations (Figure 1 - figure supplement 2d), which together represent a new intermediate along the DNA replication pathway of Bst DNAP-I.
Next, we examined whether a solution catalyzed conformation could be converted to an in crystallo conformation through a round of in crystallo catalysis. Accordingly, dATP was soaked into a crystal of the n + 1 translocated product obtained by crystallization of a solution catalyzed reaction. Following one cycle of in crystallo catalysis, an n + 2 translocated structure was produced that now contained the pre-insertion site and matched the active site conformation of previous in crystallo results (Figure 1 - figure supplement 2e, f). This observation demonstrates that in crystallo catalysis favors an active site conformation that contains the pre-insertion site, as the same active site conformation is obtained from two different starting points.
Interestingly, the translocated product obtained from the set of solution catalyzed reactions is similar to known Bst DNAP-I structures solved with duplexes that contain damaged DNA intermediates and active site mutations (Figure 1—figure supplement 5, Supplementary file 1b). These structures were previously thought to contain a distorted active site conformation due to the position of Tyr714 relative to its conformation in the in crystallo catalysis structures (Gehrke et al., 2013; Johnson and Beese, 2004; Wang et al., 2012). However, given the homology of these structures to the translocated product of solution catalyzed reactions, we postulate that Tyr714 functions as a regulatory checkpoint in the mechanism of DNA synthesis by evaluating the geometry of the newly formed base pair.
Next, we wondered whether the mechanism of DNAP-I included the formation of a pre-insertion complex, which is a ternary structure different from the previously discussed pre-insertion site observed in the binary structure of in crystallo catalyzed primer-extension reactions. Previously, Wu and colleagues solved the ternary structure of a mutant version of Bst DNAP-I bound to an incoming dATP substrate (Miller et al., 2015). Although that structure was originally described as an open ternary complex, presumably to avoid confusion with the pre-insertion site, it resembles the pre-insertion complex first observed in Klentaq1 (Li et al., 1998). The key difference between the open ternary and pre-insertion complex is whether the incoming nucleotide is paired opposite the templating base or an active site residue (Doublié et al., 1998; Yin and Steitz, 2004). Since the structure by Wu and colleagues shows the incoming substrate paired opposite Tyr714, it should be considered a pre-insertion complex.
We demonstrated that the wild-type polymerase is also capable of forming a pre-insertion complex by solving the ternary structure of the enzyme bound to the non-hydrolyzable analog, dAMPNPP. The resulting structure (Figure 2) closely resembles the mutant Bst polymerase structure determined by Wu and colleagues and shows Tyr714 paired opposite the incoming nucleotide (Miller et al., 2015). Although the phosphate tail shows nearly 100% occupancy, the sugar and nucleobase moieties are flexible, which is consistent with the dynamic properties of the incoming nucleotide in an open polymerase conformation. Nevertheless, the structure shows that the incoming nucleotide is stabilized by polar contacts to the negatively charged triphosphate moiety. These observations demonstrate that Bst DNAP-I adopts a pre-insertion complex similar to other A-family DNA polymerases (Rothwell and Waksman, 2005), which clarifies an important step in the mechanism of DNA synthesis.
Based on the structures reported here, we propose a revised mechanism for DNA synthesis by DNA polymerase I. The catalytic cycle consists of four key steps that derive from high resolution structures of Bst DNAP-I and its homolog T7 RNA polymerase (Figure 3). Starting from the newly determined post-translocation complex, the polymerase undergoes a conformation change to adopt the pre-insertion complex with an incoming nucleotide paired opposite Tyr714 in the enzyme active site. This conformational change involves release of the n + 1 templating base from its stacking interaction with Tyr719 in the O1 helix and the repositioning of Tyr714 in the enzyme active site. The enzyme then undergoes a more significant conformational change to adopt the closed ternary complex (Johnson et al., 2003), which defines the pre-catalytic state of the enzyme. Immediately following phosphodiester bond formation, the enzyme adopts a post-catalytic complex in which the primer has been extended by one nucleotide (Yin and Steitz, 2004). The enzyme then translocates to the next position on the template to initiate another cycle of nucleotide addition.
In summary, we present crystal structures of DNA polymerase I that capture the translocation and nucleotide pre-insertion steps in the DNA synthesis pathway. We suggest that these new structures, along with previously solved structures obtained by in crystallo catalysis, highlight the dynamic nature of the finger subdomain in the enzyme active site. Together, the new and existing structures expand our understanding of the mechanism of DNA synthesis by capturing important intermediates in a complicated reaction pathway.
