Magnesium Induced Assembly of a Complete DNA Polymerase Catalytic Complex

Summary

The molecular details of the nucleotidyl transferase reaction have remained speculative, as strategies to trap catalytic intermediates for structure determination utilize substrates lacking the primer terminus 3’–OH and catalytic Mg²⁺ resulting in an incomplete and distorted active site geometry. Since the geometric arrangement of these essential atoms will impact chemistry, structural insight into fidelity strategies has been hampered. Here we present the first crystal structure of a pre-catalytic complex of a DNA polymerase with bound substrates that include the primer 3’–OH and catalytic Mg²⁺. This catalytic intermediate was trapped with a non-hydrolyzable deoxynucleotide analogue. Comparison with two new structures of DNA polymerase β lacking the 3’–OH or catalytic Mg²⁺ is described. These structures provide direct evidence that both atoms are required to achieve a proper geometry necessary for an in-line nucleophilic attack of O3’ on the αP of the incoming nucleotide.

Introduction

DNA polymerases must select and incorporate a complementary deoxynucleotide from a pool of structurally similar molecules to preserve the integrity of the genome. Failure to preserve Watson-Crick base pairing rules usually results in deleterious biological consequences. Although the fidelities of DNA polymerases vary widely, the catalytic efficiencies by which they insert incorrect nucleotides is only weakly dependent on the identity of the polymerase. These differences in fidelities are due to the divergent abilities of polymerases to insert the right nucleotide (Beard et al., 2002); low fidelity enzymes insert the correct nucleotide slowly, whereas high fidelity DNA polymerases insert the correct nucleotide rapidly (Beard and Wilson, 2003).

The structural characterization of DNA polymerases from different families indicates that these enzymes have a modular domain and subdomain organization (Beard and Wilson, 2003, Steitz, 1999). The polymerase domain is generally composed of 3 subdomains with distinct functional roles. The catalytic subdomain, also referred to as the C–subdomain or palm, contributes two-structurally conserved aspartate residues that coordinate Mg²⁺. Based on the mechanism for the 3’–5’ exonuclease of Klenow fragment, a ‘two-metal ion’ mechanism for nucleotidyl transfer has been proposed (Beese and Steitz, 1991, Steitz et al., 1994). The catalytic metal (metal A) is thought to lower the pK_a of the 3’-OH of the growing primer terminus while the nucleotide binding metal (metal B) coordinates the triphosphate moiety hastening binding of the incoming nucleotide. Additionally, metal B assists PP_i dissociation. Both metals are believed to stabilize the proposed penta-coordinated transition state of the nucleotidyl transferase reaction.

DNA polymerase (pol) β belongs to the X-family of DNA polymerases and is a repair enzyme that has a key role in the base excision repair of simple DNA lesions (Beard and Wilson, 2000). Due to its small size (39 kDa), lack of a proofreading exonuclease, and many structural, kinetic, and mechanistic features it shares with other DNA polymerases, it is an ideal model system for structural studies. Although pol β exhibits unique structural attributes that reflect its cellular role, the arrangement of active site residues and substrates are remarkably similar to other DNA polymerases (Steitz et al., 1994, Steitz, 1999). Structures of pol β exist for the apoenzyme (Sawaya et al., 1994) and with non-gapped DNA (Pelletier et al., 1994), gapped DNA (Sawaya et al., 1997), lesion-containing DNA (Krahn et al., 2003, Prasad et al., 2005), mismatched DNA in the active site (Krahn et al., 2004), and mismatched DNA at the primer terminus (Batra et al., 2005). These various structures provide a context to interpret structural differences imposed by alternate ligands.

The structures of ternary substrate complexes (pol/DNA/dNTP) have been solved for members of several polymerase families providing valuable insights into substrate selection and chemistry (Table 1). However, these catalytic intermediates were trapped for structure determination by using a dideoxy-terminated primer (lacks O3’) or by using a divalent metal, such as Ca²⁺, that does not support insertion. In each case, the active site geometry for the catalytic metal is distorted owing in part to the longer coordination distances induced with a larger metal such as Ca²⁺, or the removal of the catalytic nucleophile (O3’), an atom postulated to provide an important ligand for the catalytic metal (Steitz et al., 1993). The distorted geometries of these pre-catalytic complexes have hindered detailed analyses of active site geometries as it pertains to the influence of aberrant DNA (lesions and mismatches) or nucleotide (right versus wrong) substrates on catalytic function. Furthermore, the lack of reference structures exhibiting good active site geometry for DNA polymerases from different polymerase families precludes detailed conclusions regarding structure and chemistry as it relates to function.

Table 1.

Summary of the Metal Active Site Coordination Sphere of DNA Polymerase Undamaged DNA/d(d)NTP Ternary Complex Structures^a

		Catalytic (A)			Nucleotide (B)
DNA Polymerase (Family)	NTP	Me	xb	nc	Me	x	n	Resolution	PDB ID (Ref.)
			Å			Å		Å
Klentaq (A)	ddCTP	Mg	2.37 (0.07)	5	Mg	2.22 (0.04)	6	2.30	3KTQ (Li et al., 1998)
	ddGTP	Mg	2.46 (0.09)	5	Mg	2.33 (0.06)	6	2.30	1QSS (Li et al., 1999)
	ddATP	Mg	2.48 (0.03)	5	Mg	2.21 (0.05)	6	2.30	1QSY (Li et al., 1999)
	ddTTP	Mg	2.44 (0.05)	5	Mg	2.24 (0.03)	6	2.30	1QTM (Li et al., 1999)
T7 (A)	ddGTP	Mg	2.40 (0.04)	5	Mg	2.27 (0.03)	6	2.20	1T7P (Doublié et al., 1998)
	ddATP	Mg	2.58 (0.06)	5	Mg	2.29 (0.03)	6	2.40	1SKR (Li et al., 2004)
	ddCTP	Mg	2.52 (0.09)	5	Mg	2.34 (0.04)	6	2.54	1T8E (Brieba et al., 2004)
BF (A)	dCTP	Mg	2.75 (0.09)	3	Mn	2.23 (0.12)	6	1.95	1LV5 (Johnson et al., 2003)
RB69 (B)	dTTP	Ca	3.08 (0.09)	4	Ca	2.66 (0.02)	7	2.60	1IG9 (Franklin et al., 2001)
HIV-1 RT (RT)	dTTP	Mg	2.71 (0.21)	3	Mg	2.36 (0.07)	6	3.20	1RTD (Huang et al., 1998)
β (X)	ddCTP	Mgd	2.35 (0.01)	4	Mg	2.00 (0.03)	6	2.20	1BPY (Sawaya et al., 1997)
	dTMPPCP	–e	–	–	Cr	2.15 (0.05)	5	2.60	1HUO (Arndt et al., 2001)
λ (X)	ddTTP	–	–	–	Mg	2.06 (0.05)	6	1.95	1XSN (Garcia-Diaz et al., 2005)
Dpo4 (Y)	ddGTP	Mg	2.41 (0.20)	4	Ca	2.34 (0.04)	6	2.10	1JXL (Ling et al., 2001)
τ (Y)	dTTP	–	–	–	Mg	2.41 (0.08)	6	2.30	1ZET (Wang, 2005)
τ (Y)	dCTP	Mg	2.66 (0.20)	4	Mg	2.21 (0.04)	6	2.50	1ALZ (Nair et al., 2005b)
REV1 (Y)	dCTP	Mg	2.57 (0.06)	4	Mg	2.28 (0.05)	6	2.30	2AQ4 (Nair et al.,

^aExcept for the Dpo4 and pol β(dTMPPCP) structures, the 3’-primer termini in these structures lack O3’ (dideoxy-terminated).

^bAverage coordination distance (SE).

^cCoordination number (oxygen ligands <3.5 Å). In some instances, both oxygens of the active site aspartates were near metal A or B. In these cases, the closest oxygen was assumed to be the preferred ligand.

^dThe identity of this ion is now believed to be Na⁺ (Figure 2).

^eMetal not observed.

By using a non-hydrolysable dUTP analogue, we have determined the structure of a pre-catalytic complex of pol β that includes the catalytic Mg²⁺ and O3’ of the primer terminus. For the first time, there is direct evidence that the catalytic Mg²⁺ coordinates the primer O3’, induces subtle conformational rearrangements resulting in good octahedral geometry, and positions O3’ for an in-line nucleophilic attach on αP of the incoming nucleotide. By employing a similar approach with alternate substrates or DNA polymerases, similarities and/or differences in active site geometries can be reliably assessed.

Results and Discussion

High Resolution Ternary Substrate Complex with Dideoxy-terminated Primer

Using the strategy of employing a dideoxy-terminated primer to trap a ternary substrate complex, previously reported structures of pol β failed to reveal the complete inner coordination sphere of the catalytic metal (Pelletier et al., 1994, Sawaya et al., 1997). We have collected new data and determined a high-resolution (1.65 Å) structure of pol β (Table 2). The new structure provides important details related to the coordination and identity of the ion bound in the catalytic metal site. A water molecule that had been suggested to participate in the octahedral coordination sphere of the catalytic magnesium can now be observed (S2 in Figure 1A). The sixth coordinating ligand is proposed to be O3’ of the primer terminus that is missing in this structure (see below). Additionally, the sugar pucker of the dideoxy-terminated primer is C3’-endo, similar to that observed in the well-defined structure of the ternary complex of Bacillus DNA polymerase I fragment (Johnson et al., 2003), an A-family member, and a conformation commonly observed in A-form DNA.

Table 2.

Crystallographic Statistics

Complexa A–B	ddCTP Na+–Mg2+	dUMPNPP Na+–Mg2+	dUMPNPP Mg2+–Mg2+
Data Collection
a (Å)	50.9	50.8	50.8
b (Å)	80.4	80.2	80.4
c (Å)	55.3	55.3	55.5
β (°)	107.2	107.5	107.9
dmin(Å)	1.65 (1.71–1.65)b	2.20 (2.28–2.20)	2.00 (2.07–2.00)
Rmerge(%)c	4.1 (24.9)d	7.2 (43.2)	6.4 (50.1)
Completeness (%)	97.5 (86.5)	98.8 (97.2)	99.7 (97.9)
I/σI	22.8 (4.2)	15.0 (3.8)	10.6 (2.0)
Number of observed reflections	349479	123108	118197
Number of unique reflections	50030	21443	28459
Radiation source	Rigaku RUH3R	Rigaku RUH3R	APS
Wavelength	1.5418	1.5418	0.9997
Refinement
Rms deviations
Bond lengths (Å)	0.007	0.005	0.005
Bond angles (°)	1.020	1.050	1.050
Rwork(%)e	18.7	19.0	19.6
Rfree(%)f	21.9	24.3	24.4
Average B factor (Å2)
Protein	20.0	30.0	34.1
DNA	24.8	37.0	38.3
Metal A	15.0	29.6	21.5
Metal B	10.6	18.3	17.5
Ramachandran analysis (%)g
Favored	97.8	98.8	98.2
Allowed	100.0	100.0	100.0

^aRefers to the incoming nucleoside triphosphate of the ternary substrate complex. The metals refer to those identified in the catalytic (A) and nucleotide metal (B) binding sites.

^bValues in parentheses refer to the highest resolution shell.

^cR_merge = 100 × ∑_h ∑_i |I_h,i −I_h| ∑_h ∑_i I_h,i, where I_h is the mean intensity of symmetry-related reflections I_h,i.

^dNumbers in parentheses refer to the highest resolution shell of data (10%).

^eR_work = 100 × ∑||F_obs| −|F_calc||/∑|F_obs|

^fR_free for a 5% subset of reflections withheld from refinement.

^gAs determined by MolProbity (Lovell et al., 2003).

Figure 1. Active Site Coordination of the Ions Observed in a High-Resolution (1.65 Å) DNA Polymerase β /DNA/ddCTP Teranry Complex.

(A) A F_o-F_c simulated annealing electron density omit map (gray) contoured at 2.5σ for the pol β active site. A penta-coordinated ion is bound to the catalytic metal site. The coordination distances (Table 3) are consistent with the identity of this ion as Na⁺ (light blue). Additionally, a previously postulated water molecule (S2) is now observed. The Na⁺ is also coordinated by all three active site aspartates (D190, D192, D256) and a non-bridging oxygen on Pα (pro-R_P oxygen). The nucleotide binding metal is consistent with Mg²⁺ (orange). The density is superimposed on the refined modeled primer terminus (P10), incoming nucleotide (ddCTP), active site aspartates (D190, D192, D256), and a nucleotide metal-coordinating water. The primer terminus O3’ is believed to be the sixth ligand, but was omitted to trap the catalytic intermediate for structure determination.

(B) The octahedral coordination and coordination distances (Table 3) of the ion in the nucleotide binding metal site is consistent with the identity of this ion as Mg²⁺. The metal is coordinated to non-bridging oxygens on all three phosphates of the incoming ddCTP (α,β ,γ-tridentate). In addition, Asp190 (D190), Asp192 (D192), and a water molecule (S1) completes the inner coordination sphere.

Most importantly, however, the ion occupying the catalytic metal site in this high-resolution structure appears to be Na⁺ rather than Mg²⁺. Since Mg²⁺ has a low electron density and has the same number of electrons as Na⁺ and water, it is difficult to unambiguously assign magnesium ions in protein crystal structures. The identity of this ion is based on the average coordination distance of the coordinating ligands (2.38 Å; Figure 2 and Table 3). This distance is similar to that expected of Na⁺ (2.42 Å) (Harding, 2002) rather than Mg²⁺ (2.07 Å) (Harding, 2001), observed in accurately determined small molecule structures (Cambridge Structural Database). Indeed, the average coordination distance observed for the six inner sphere oxygen ligands of the ion in the nucleotide metal B site is consistent with Mg²⁺ occupying this site (2.08 Å, Figures 1B and 2). The penta-coordination geometry of the cation occupying the catalytic metal site is distorted, presumably due to the lack of one of the coordinating ligands (primer terminus O3’). Further, employing the bond-valence method is consistent with the ion occupying the catalytic metal site as Na⁺ rather than Mg²⁺ (Müller et al., 2003). The average coordination distance observed in our previously reported ternary complex structure of pol β is also consistent with Na⁺ occupying the catalytic metal site, Figure 2 (Sawaya et al., 1997).

Figure 2.

Average Coordination Distances for the Oxygen Ligands to the Catalytic (A) or Nucleotide Binding (B) Metals

These distances are computed for a previously reported structure of pol β (1BPY) (Sawaya et al., 1997), the 1.65 Å structure (ddCTP/Na–Mg; 2FMP), and two ternary substrate complex structures trapped with the dUMPNPP non-hydrolyzable analogue (Na–Mg, 2FMQ; Mg–Mg, 2FMS). The substrate complex structures are denoted by the identity of the incoming nucleotide and the putative ions in sites A and B. The specific coordinating atoms and distances are tabulated in Table 3. The errors bars represent the standard errors of the calculated averages. The horizontal dashed lines indicate the coordinate distances observed from accurately determined small molecule structures (Harding, 2001, Harding, 2002).

Table 3.

Active Site Metal Coordination Distances (Å)

Ternary Complexa A–B	ddCTPb Na+–Mg2+	ddCTP Na+–Mg2+	dUMPNPP Na+–Mg2+	dUMPNPP Mg2+–Mg2+
Metal A
P10 (O3’)	—c	—c	3.5	2.2
D190 (OD2)	2.2	2.3	2.6	2.1
D192 (OD1)	2.0	2.2	2.2	2.2
D256 (OD2)	2.6	2.4	2.8	2.1
NTP (O1A)	2.4	2.5	2.5	2.2
H2OS2(O)	—d	2.5	—d	2.2
Metal B
NTP (O1A)	1.9	2.1	2.1	2.1
NTP (O2B)	2.1	2.0	2.0	2.0
NTP (O3G)	2.0	2.1e	2.1	2.1
D190 (OD1)	2.0	2.1	2.1	1.9
D192 (OD2)	2.0	2.0	2.0	2.0
H2OS1 (O)	2.0	2.2	2.1	2.3

^aRefers to the incoming nucleoside triphosphate of the ternary substrate complex. The metals refer to those identified in the catalytic and nucleotide metal binding sites; metal A and B, respectively.

^bFrom Sawaya et al. (Sawaya et al., 1997); pdb accession code 1BPY. Metal A was originally identified as Mg²⁺, but the coordination distances are more consistent with Na⁺ occupying this site.

^cThe sugar at the primer terminus lacks a 3’-OH.

^dThis solvent molecule was not observed.

^ePγ exhibited 50% occupancy; accordingly, this distance represents the average distance between O3G and a water molecule at this position.

Ternary Substrate Complex with Primer 3’-OH

The catalytic metal sites of most previously determined substrate pre-catalytic complex structures of DNA polymerases are missing a critical coordinating ligand, O3’ of the primer terminus (Table 1). Since 3’-O⁻ serves as the attacking nucleophile, the position and coordination geometry is expected to have important ramifications for transition state development. Accordingly, we have determined a structure of a ternary complex that includes 3’–OH of the primer terminus. This was achieved by using a commercially available dNTP analogue, 2’-deoxy-uridine-5’-(α,β)-imido triphosphate (dUMPNPP). The bridging oxygen between the α- and β-phosphates is replaced with nitrogen in this dUTP analogue which prevents insertion. DNA polymerases efficiently insert dUTP since it closely resembles dTTP and it does not alter Watson-Crick hydrogen bonding. DNA polymerase β inserts dUTP with the same efficiency as dTTP (Figure 3A) and is competitively inhibited by dUMPNPP (K_i = 0.9 μM; Figure 3B).

Figure 3. Kinetic Analysis of dUTP Insertion and dUMPNPP Inhibiiton.

(A) Steady-state kinetic analysis indicates that dUTP insertion (■) is similar to that for dTTP (●). The catalytic efficiencies (k_cat/K_m) for dTTP and dUTP insertion are 0.19 ± 0.03 and 0.15 ± 0.02 1/μM–s, respectively, for at least two independent determinations.

(B) Dixon analysis for competitive inhibition indicates that the K_i for dUMPNPP is 0.9 μM (dotted line).

A crystal of the binary single-nucleotide gapped DNA complex was grown and the nucleotide analogue soaked into the crystal (Batra et al., 2005). This crystal diffracted at 2.2 Å, and clear density for the O3’ on the primer terminus was observed (Figure 4A). In addition, replacing the bridging oxygen between the α- and β-phosphates with nitrogen did not significantly distort the triphosphate moiety of the incoming nucleotide as evidenced by the equivalent position of the phosphates for ddCTP and dUMPNPP (Figure 4B). Unexpectedly however, the primer terminal O3’ was 4.7 Å from the αP of the incoming dUMPNPP and 3.5 Å from the ion in catalytic metal site (Figure 4C). Despite the presence of magnesium in the crystallization solution, the ion in the catalytic metal site was consistent with Na⁺ rather than Mg²⁺ (Figure 2). Here, as in the high-resolution structure described above, the identity of the nucleotide binding metal B was consistent with Mg²⁺ displaying good octahedral geometry and an average coordination distance of 2.07 Å (Figures 2 and 4B).

Figure 4. Active Site Geometry of the Ternary DNA Polymerase β /DNA/dUMPNPP Complex with Na⁺ in the Catalytic Metal Site.

(A) The F_o-F_c simulated annealing electron density omit map (gray) contoured at 3.8σ for the pol β active site with Na⁺ (light blue) occupying the catalytic site. The nucleotide binding metal is consistent with Mg²⁺ (orange). The density is superimposed on the refined modeled primer terminus (P10), incoming nucleotide analogue (dUMPNPP), active site aspartates (D190, D192, D256), and the nucleotide metal-coordinating water (S1). The density for O3’ is clearly observed but distant from the sodium ion in the catalytic metal site.

(B) The ternary substrate complex structures of pol β with an incoming ddCTP (yellow carbon and orange phosphates atoms) and dUMPNPP (green carbon and magenta phosphate atoms) were superimposed with the catalytic subdomains (RMSD = 0.13 Å for 344 atom pairs). The superimposed structures clearly indicate that the bridging nitrogen (blue) between Pα and Pβ does not significantly alter the position of the phosphates. The octahedral coordination and coordination distances (Table 3) of the ion in the nucleotide binding metal site is consistent with the identity of this ion as Mg²⁺.

(C) The Na⁺ (light blue) occupying the catalytic metal site displays a distorted coordination geometry. The four coordinating oxygens are displayed as red balls. The primer terminus (P10) O3’ is too distant (3.5 Å) for Na⁺ to participate in its coordination. Furthermore, O3’ is 4.7 Å from Pα of the incoming nucleotide analog.

Pre-catalytic Complex with Mg²⁺ Occupying the Catalytic Metal Site

The nucleotide analogue had been soaked into the binary complex crystals in the presence of 180 mM sodium acetate and 10 mM MgCl₂. To provide Mg²⁺ a better opportunity to bind to the catalytic metal site, the Mg²⁺/Na⁺ ratio was increased from 0.06 to 2.22 (see Experimental Procedures). The resulting crystal diffracted at 2.0 Å and there is clear density for O3’ of the primer terminus (Figure 5). The octahedral coordination geometry (Figure 6A) and average coordination distance for the oxygen ligands with the ion in site A was consistent with Mg²⁺ occupying this site (Figures 2). Significantly, O3’ of the primer terminus is now 2.2 Å from metal A in contrast to the 3.5 Å observed when Na⁺ occupies this site (Figure 7). This repositioning of O3’ could occur through a change in the preferred sugar-pucker of the primer terminus. The sugar-pucker of the primer terminus with Mg²⁺ occupying the catalytic metal site is C3’-endo (phase angle of pseudorotation, P = 29°). The resultant sugar conformation positions O3’ in a near perfect geometry for an in-line attack on the αP of the incoming nucleoside triphosphate. This new position not only completes the octahedral inner coordination sphere for the catalytic Mg²⁺ (Figure 6A), but also situates O3’ of the primer terminus within 3.4 Å of the αP of the incoming dCTP (Figure 7). The nucleotide binding Mg²⁺ in site B coordinates nonbridging oxygens from each phosphate (α,β,γ–tridentate) and exhibits good octahedral geometry when site A is occupied by either Na⁺ or Mg²⁺ (Figures 4B and 6B).

Figure 5. Active Site Geometry of the Ternary DNA Polymerase β /DNA/dUMPNPP Complex with Mg²⁺ in the Catalytic Metal Site.

The F_o-F_c simulated annealing electron density omit map (gray) contoured at 4.7σ for the pol β active site with Mg²⁺ (orange) occupying both the catalytic and nucleotide metal binding sites. The density is superimposed on the refined modeled primer terminus (P10), incoming nucleotide analogue (dUMPNPP), active site aspartates (D190, D192, D256), and the metal-coordinating waters (S1 and S2). The density for O3’ is clearly observed and the resulting C3’-endo sugar pucker positions O3’ near the catalytic metal and Pα of dUMPNPP.

Figure 6. Coordination Geometry of the Metal Binding Sites in the Complete Pre-Catalytic Complex.

(A) Coordination geometry of Mg²⁺ (orange) occupying the catalytic metal site. The sugar pucker of the primer terminus (P10) is C3’-endo and the catalytic metal exhibits good octahedral geometry. Additionally, a water molecule (S2) not observed in the lower resolution Na⁺ structure is visible and participates in the first coordination sphere of the catalytic Mg²⁺.

(B) The Mg²⁺ occupying the nucleotide metal site also exhibits good octahedral geometry even when the catalytic site is occupied by Na⁺ (Figure 4B).

Figure 7. Stereo View of the Pol β Active Site.

Superimposed structures of pol β with either Na⁺ (light blue) or Mg²⁺ (yellow) in the catalytic metal site A. Residues that hydrogen bond (green) to the triphosphate moiety of the incoming nucleotide and the active site aspartates are shown. When Na⁺ occupies the catalytic metal site, O3’ of the primer terminus is 3.5 and 4.7 Å from the catalytic metal and αP of the incoming nucleotide, respectively. In contrast, with Mg²⁺ in the catalytic metal site, an altered sugar pucker positions O3’ of the primer terminus 2.2 and 3.4 Å from the catalytic metal and Pα of the incoming nucleotide, respectively.

Polymerase Metal Coordinating Ligands

Three acidic residues of the catalytic subdomain coordinate the two metals in the ternary substrate complex: Asp190, Asp192, and Asp256 (Figures 6 and 7). Asp190 and Asp192 coordinate both metals and are structurally conserved among members of different polymerase families. In contrast, Asp256 only coordinates metal A, and the equivalent residue in A– and Y–family polymerases is glutamate. Structural alignment of substrate complexes from members of these families, based on the triphosphate moiety of the incoming nucleotide, indicates that these glutamate residues could coordinate a catalytic Mg²⁺ in an alternate conformation. The proposed role of Asp256, or its acidic counterpart in other polymerases, as a general base is controversial (Pelletier, 1994, Steitz et al., 1994). The geometry and position of the catalytic metal provides compelling evidence that it can alter the pK_a of both the primer terminal 3’–OH and Asp256 (Pelletier, 1994). Alanine-substitution for Asp256 abolishes pol β activity demonstrating a critical role of this residue during catalysis (Sobol et al., 2000). Likewise, alanine-substitution for Glu883 in Klenow fragment of E. coli DNA polymerase I dramatically decreases the rate of chemistry indicating an important role of this residue in A-family DNA polymerases (Polesky et al., 1992). Since the nature of the metal at site A has a significant influence on the geometry and position of the surrounding oxygen ligands, the identity of the metal occupying the catalytic site is crucial to a structural interpretation of the role of this acidic side chain. In this regard, we note that the average coordinate distance of the ligands for the metal modeled in site A in other reported DNA polymerase structures is greater than 2.35 Å and the coordination number less than that expected for magnesium (n=6) (Table 1). It remains to be determined whether these differences represent alternate cations bound to these sites or if it represent intrinsic differences that may have functional implications.

In the absence of the incoming nucleotide (pol βDNA binary complex), neither metal site is occupied and Asp190, Asp192, and Asp256 are observed in alternate conformations, hydrogen bonding to the primer terminal O3’, Arg258, and Arg254, respectively (Sawaya et al., 1997). Additionally, the binding of the correct nucleotide, with its associated Mg²⁺ (metal B), induces subdomain motions and DNA conformational changes that ‘close’ the nascent base pair pocket necessary for efficient DNA synthesis (Beard and Wilson, 2003). Thus, dNTP/metal B and metal A binding result in crucial conformational changes in the active site. Even subtle alterations in the active site by exchanging the cation in metal site A demonstrates that the polymerase active site is exquisitely sensitive to alternate cations at this site. For example, the conformation of Asp190 and Asp256 and the position of O3′ of the primer terminus are dependent on the identity of the metal occupying this site. Yet, other residues in the vicinity of the incoming nucleotide (e.g., Arg183) are not sensitive to the identity of the metal in site A (Figure 7). Using an exchange inert Cr³⁺—dNTP complex, a ternary substrate complex structure of pol β was determined where the catalytic metal site is deliberately left empty so as to trap a the complex where the primer terminus had a 2′-deoxyribose sugar, Table 1 (Arndt et al., 2001). In this situation, Asp256 displays yet another conformation pointing away from the empty catalytic metal binding site.

Concluding Remarks

Based on a two metal mechanism proposed for the 3′–5′ exonuclease activity of the Klenow fragment of E. coli DNA polymerase I, Beese and Steitz (Beese and Steitz, 1991) suggested that polymerases might utilize a similar mechanism. Subsequently, the first ternary substrate complex of a DNA polymerase was solved by employing a dideoxy-terminated primer that captured a ternary pol β/DNA/ddCTP complex (Pelletier et al., 1994). Many aspects of this structure support a two-metal mechanism; particularly the observation that active site acidic groups coordinate two divalent metal which are approximately positioned to participate in chemistry. Although magnesium was modeled into the catalytic metal site in that early structure, it was noted that the electron density could have been interpreted as a water molecule. Subsequent ternary complex structures have trapped catalytic intermediates by utilizing a dideoxy-terminated primer (Table 1). In many of these cases, a magnesium ion was modeled into the catalytic metal site even though the number of coordinating ligands and distances were not appropriate for magnesium (Harding, 2001). The lack of a potential coordinating ligand (O3′) would be expected to reduce the affinity and specificity of metal site A, but not site B. In this regard, substrate structures of pol β indicate that Mg²⁺ occupies site B in the presence of an incoming nucleotide regardless whether the incoming nucleotide is ddCTP or dUMPNPP (Figures 1B, 2 and 4B). Likewise, ternary complex structures of pol β determined in the presence of high sodium concentrations indicate that the catalytic metal site is occupied by Na⁺, rather than Mg²⁺, independent of the identity of the incoming nucleotide (ddCTP and dUMPNPP) or whether the primer terminus has a 3′-OH (Figures 1B, 2, and 4B). Since the intracellular concentration of sodium is low, this observation is probably not physiologically relevant. However, the results indicate that when ions other than magnesium are bound to the catalytic metal site, the coordination and active site geometry (O3′–Pα distance) are distorted (Figure 7). We have not tried to increase the Mg²⁺ concentration with the ddCTP closed complex to see whether Mg²⁺ can replace Na⁺ in the catalytic metal site.

Nucleophilic substitution reactions at phosphorous are portrayed relative to bond formation/breaking character at the transition state and are expressed in terms of intermediates between associative and dissociative mechanisms. A minimum estimate of the fractional associativity of a mechanism can be calculated from the distance between the attacking atom (3′-O⁻) and the phosphorous undergoing substitution (Pα) from a ground-state crystal structure (Mildvan, 1997). Assuming a symmetrical trigonal-bipyramidal transition state where the atoms occupying the apical positions (O3′ and NH^α–β) are stationary, the distance between O3′ and Pα would be 2.5 Å, indicating a lower limit of 5.2% associativety for this mechanism.

Kinetic analyses of DNA polymerases from various polymerase families suggest that under some conditions non-chemical steps precede nucleotidyl transfer and limit correct nucleotide insertion (Joyce and Benkovic, 2004). The identity of this step(s) in terms of structural events is unknown. Although there is no definitive evidence indicating that a non-chemical step limits correct nucleotide insertion for pol β, structural (Sawaya et al., 1997) and kinetic studies (Beard et al., 2004) clearly demonstrate that conformational changes occur upon binding substrates for pol β. These events include subdomain motions (Sawaya et al., 1997), sequential protein side-chain rearrangements (Yang et al., 2002), and DNA conformational changes that include template reorientation (Beard and Wilson, 1998). The thermodynamic contribution of these observed conformational changes to the overall fidelity of DNA synthesis remains to be established. A simple change in the sugar pucker of the primer terminus to C3′-endo is sufficient to position O3′ near Pα (Figure 7). Although alterations in sugar pucker are generally very rapid in solution, it remains to be determined what parameters in the confines of the polymerase active site influence the observed sugar conformation. We demonstrate here that the identity of the ion in site A influences the primer terminus sugar pucker: C3′-endo (P = 29°) in the presence of Mg²⁺ and P ≫ 29° when Na⁺ occupies this site. The resolution of the Na⁺ complex with dUMPNPP (2.2 Å) precludes unambiguous description of the sugar pucker of the primer terminus. However, the positions of O3′ is clear and indicates that the conformation should be nearer C2′-endo (Figures 4A, 5, and 7). Thus, in addition to the multitude of conformational transitions described previously (Beard and Wilson, 1998, Beard and Wilson, 2003, Pelletier et al., 1994, Sawaya et al., 1997), catalytic magnesium binding induces additional conformational adjustments that bring about proper geometrical alignment necessary for efficient nucleotidyl transfer.

The use of substrate analogs to trap catalytic intermediates for structure determination has been a useful and productive approach. The use of a DNA analog lacking a primer 3′-OH has been used successfully many times (Table 1) since its first application with pol β (Pelletier et al., 1994). We now demonstrate the utility of using a non-hydrolyzable dNTP analog. This approach captures a catalytic intermediate without removing any of the atoms that participate directly during catalysis (primer-O3′, Pα, and Mg²⁺). Instead, the bridging oxygen between Pα and Pβ is substituted with nitrogen. The nitrogen substitution alters the PA–X3A bond distance and PA–X–PB angle less than 7 and 3% respectively, thereby preserving active site geometry.

The structures reported here reveal the dynamic nature of the polymerase active site (Bose-Basu et al., 2004) as well as the strong influence of the catalytic metal on active site geometry (Yang et al., 2004). The aberrant conformations of mismatched base pairs at the primer terminus (Batra et al., 2005) or nascent base pair binding pocket (Krahn et al., 2004) suggest that misinsertion is deterred by distorting the catalytic metal binding site. The structure of the complete catalytic complex with Mg²⁺ occupying both metal binding sites represents the closest structure to the transition state for a DNA polymerase to date. It is now possible to infer active site adjustments when unnatural metals that reduce fidelity, such as Co²⁺ and Mn²⁺ (Sirover and Loeb, 1976), occupy the metal binding sites. For the first time, we can now begin to confidently examine the precise sequence of events (deprotonation/protonation and bond making/breaking) and the free energy surface of the chemical reaction by computational methods. Finally, the use of dNTP analogs to trap pre-catalytic complexes with other DNA polymerases should uncover general and specific strategies that are utilized to achieve efficient and faithful replication of DNA.

Experimental Procedures

Crystallization of the Pol β Ternary Substrate Complex

Human pol β was over-expressed in E. coli and purified (Beard and Wilson, 1995). The DNA substrate consisted of a 16-mer template, a complementary 9- or 10-mer primer strand, and a 5-mer downstream oligonucleotide. The annealed 9-mer primer creates a two-nucleotide gap with two templating G residues. Proper annealing with the 10-mer produces a one-nucleotide gapped DNA substrate with an adenine serving as the templating base. The sequence of the downstream oligonucleotide was 5′–GTCGG–3′ with a phosphorylated 5′-terminus. The template sequence was 5′–CCGAC(A or G)GCGCATCAGC–3′ and the primer sequence was 5′–GCTGATGCG(±C)–3′. Oligonucleotides were dissolved in 20 mM MgCl₂, 100 mM Tris/HCl, pH 7.5. Each set of template, primer, and downstream-oligonucleotides was mixed in a 1:1:1 ratio and annealed using a PCR thermocycler by heating 10 min at 90 °C and cooling to 4 °C (1 °C min⁻¹) resulting in a 1 mM mixture of gapped duplex DNA. This solution was then mixed with an equal volume of pol β at 4 °C, the mixture warmed to 35 °C and gradually cooled to 4 °C. A four-fold molar excess of ddCTP was added to the 2-nucleotide gapped DNA to produce the ternary DNA/ddCTP complex.

Pol β-DNA complexes were crystallized by sitting drop vapor diffusion. The reservoir solution for formation of the ternary DNA/ddCTP complex consisted of 50 mM HEPES, pH 7.5, 180 mM sodium acetate and 16% PEG-3350. The crystallization buffers for binary one-nucleotide gapped DNA complexes were 16% PEG-3350, 350 mM sodium acetate, and 50 mM imidazole, pH 7 or 100 mM imidazole, pH 8. Drops were incubated at 18 °C and streak seeded after 1 day. Crystals grew in approximately 2 to 4 days after seeding. The crystals of the ternary DNA/ddCTP complex were transferred to cryo-protectant solutions containing 50 mM HEPES, pH 7.5, 180 mM sodium acetate, 12% ethylene glycol and 20% PEG-3350. The crystals of the binary one-nucleotide gapped DNA complex were soaked in artificial mother liquor with10 mM MgCl₂, 1 mM dUMPNPP (Jena Bioscience), 20% PEG-3350, and 12% ethylene glycol. To saturate Mg²⁺ binding to the catalytic site, the [Mg²⁺]/[Na⁺] in the soaking solution was increased by lowering the sodium acetate concentration to 90 mM, and increasing MgCl₂ to 200 mM. All three crystals belong to the space group P2₁.

Data Collection and Structure Determination

Data were collected at 100 K on an R-Axis IV area detector system mounted on a RUH3R rotating anode generator or at the Advanced Photon Source (APS, SER-CAT 22-ID). Data were integrated and reduced with HKL2000 software (Otwinowski and Minor, 1997).

Ternary substrate complex structures were determined by molecular replacement with a previously determined structure of pol β complexed with one-nucleotide gapped DNA and a complementary incoming ddCTP, PDB accession 1BPY (Sawaya et al., 1997). The crystal structures have similar lattices and are sufficiently isomorphous to determine the molecular-replacement model position by rigid body refinement. The parameters and topology for dUMPNPP were based on CNS libraries with parameters for the P–N–P fragments derived from analysis of several well-characterized small molecule structures (Cambridge Structure Database). Nucleic acid backbone and sugar pucker dihedral restraints were excluded in later stages of refinement because CNS enforces standardized A/B-form geometries, whereas several residues in these structures exhibit alternate geometries (Krahn et al., 2004). Refinement statistics are given in Table 2. The figures were prepared in Chimera (Pettersen et al., 2004) or Molscript/Raster3D (Kraulis, 1991, Merritt and Bacon, 1997).

Kinetic Assays

DNA synthesis was assayed on a single-nucleotide gapped DNA substrate where the templating base in the gap was an adenine. The DNA sequence was as described previously (Beard et al., 2004) with the core sequence identical to that used for crystallization. Enzyme activities were determined using a reaction mixture containing 50 mM Tris-HCl, pH 7.4, 20 mM MgCl₂, 200 nM single-nucleotide gapped DNA, and various dUTP or dTTP concentrations. Reactions were initiated with 0.5 nM enzyme at room temperature and stopped with EDTA mixed with formamide dye. The substrates and products were separated on 15% denaturing polyacrylamide gels and quantified in the dried gels by phosphorimagery. Steady-state kinetic parameters were determined by fitting the rate data to the Michaelis equation. Inhibition of dTTP insertion by dUMPNPP was determined at various concentrations of inhibitor and sub-saturating dTTP concentrations.