Versatility of Y-family Sulfolobus solfataricus DNA Polymerase Dpo4 in Translesion Synthesis Past Bulky N2-Alkylguanine Adducts*

In contrast to replicative DNA polymerases, Sulfolobus solfataricus Dpo4 showed a limited decrease in catalytic efficiency (kcat/Km) for insertion of dCTP opposite a series of N2-alkylguanine templates of increasing size from (methyl (Me) to (9-anthracenyl)-Me (Anth)). Fidelity was maintained with increasing size up to (2-naphthyl)-Me (Naph). The catalytic efficiency increased slightly going from the N2-NaphG to the N2-AnthG substrate, at the cost of fidelity. Pre-steady-state kinetic bursts were observed for dCTP incorporation throughout the series (N2-MeG to N2-AnthG), with a decrease in the burst amplitude and kpol, the rate of single-turnover incorporation. The pre-steady-state kinetic courses with G and all of the six N2-alkyl G adducts could be fit to a general DNA polymerase scheme to which was added an inactive complex in equilibrium with the active ternary Dpo4·DNA·dNTP complex, and only the rates of equilibrium with the inactive complex and phosphodiester bond formation were altered. Two crystal structures of Dpo4 with a template N2-NaphG (in a post-insertion register opposite a 3′-terminal C in the primer) were solved. One showed N2-NaphG in a syn conformation, with the naphthyl group located between the template and the Dpo4 “little finger” domain. The Hoogsteen face was within hydrogen bonding distance of the N4 atoms of the cytosine opposite N2-NaphG and the cytosine at the –2 position. The second structure showed N2-Naph G in an anti conformation with the primer terminus largely disordered. Collectively these results explain the versatility of Dpo4 in bypassing bulky G lesions.

Accurate and efficient replication of DNA is crucial for the preservation of genomic integrity and survival of organisms (1). A major obstacle for DNA replication is caused by various DNA lesions, which are inevitably formed by both endogenous sources and exogenous mutagens in cells and are usually present in replicating DNA after avoiding repair (2). DNA polymerases often produce abnormal phenomena when encountering lesions during DNA replication, including misinsertion of incorrect nucleotides, slippage, and blockage of replication, all of which can result in mutation or cell death and can lead to detrimental effects, including aging and cancer in mammals (3).
The nature of the mutations produced during replication is not only dependent on the specific damage in the DNA but also on the DNA polymerases involved (19,20). Replicative DNA polymerases replicate unmodified DNA bases with high fidelity and efficiency but are generally intolerant of the DNA distortions caused by many DNA lesions; thus, these polymerases are blocked and/or inefficiently misinsert bases opposite DNA adducts during replication. The "translesion synthesis" DNA polymerases have more open and larger active sites and can synthesize DNA across and beyond various replication-blocking DNA lesions (21,22). However, synthesis can still be inhibited (or blocked) by certain DNA lesions and bypass can be error-prone or error-free, depending on the adduct (23).
In the present work, Dpo4 was studied in detail regarding both kinetics and structure in the processing of a series of N 2 -guanyl adducts ranging in size, including N 2 -MeG, N 2 -EtG, N 2 -IbG, N 2 -BzG, N 2 -NaphG, and N 2 -AnthG. Two crystal structures of Dpo4 with an oligonucleotide containing an N 2 -NaphG adduct were determined. The complexes are identical in their composition but differ in the arrangement of the template N 2 -NaphG and in the pairing schemes. A kinetic model involving active and inactive DNA polymerase complexes was developed to fit the available data and is discussed in relation to the available structural data.

EXPERIMENTAL PROCEDURES
Materials-Dpo4 was expressed in E. coli and purified to electrophoretic homogeneity as described previously (46). Unlabeled dNTPs were obtained from Amersham Biosciences, (S p )-dCTP␣S and (S p )-dTTP␣S were purchased from Biolog Life Science Institute (Bremen, Germany), and [␥-32 P]ATP (specific activity 3 ϫ 10 3 Ci mmol Ϫ1 ) was from PerkinElmer Life Sciences. T4 polynucleotide kinase and restriction endonucleases were purchased from New England Biolabs (Beverly, MA). Bio-spin columns were obtained from Bio-Rad (Hercules, CA).
Reaction Conditions for Enzyme Assays and Analysis-Standard DNA polymerase reactions were carried out in 50 mM Tris-HCl buffer (pH 7.5 at 25°C) containing 50 mM NaCl, 5 mM dithiothreitol, 100 g of bovine serum albumin ml Ϫ1 , and 5% (v/v) glycerol) at 37°C (24,(27)(28)(29)48). Primer ( 32 P-labeled at the 5Ј-end using T4 polynucleotide kinase/[␥-32 P]ATP) was annealed to either an unmodified or modified (N 2 -alkyl-G) template by heating the primer and template (equimolar mixture) to 95°C for 5 min and then slowly cooling to room temperature. All reactions were initiated by the addition of dNTP solutions containing MgCl 2 (5 mM final concentration) to the preincubated enzyme/DNA mixtures. After reaction, 5-l aliquots of the reaction mixture were quenched with EDTAformamide solution (50 l of 20 mM EDTA in 95% formamide (v/v) with 0.5% bromphenol blue (w/v) and 0.05% xylene cyanol (w/v)). Products were resolved using a 20% polyacrylamide (w/v) denaturing gel electrophoresis system containing 8 M urea and then visualized and quantitated by PhosphorImager analysis using a Bio-Rad Molecular Imager FX instrument and Quantity One software (Bio-Rad).

Function of Dpo4 with N 2 -Guanyl Adducts
or 10 nM) in 50 mM Tris-HCl buffer (pH 7.5 at 25°C) containing 50 mM NaCl, 5 mM dithiothreitol, 100 mg of bovine serum albumin ml Ϫ1 , and 5% glycerol (v/v)). Reactions were initiated by adding dNTP⅐Mg 2ϩ (100 M each dNTP and 5 mM MgCl 2 ) solution at 37°C. At the indicated time points, reactions were quenched by the addition of EDTA-formamide solution and analyzed by gel electrophoresis.
Steady-state Kinetic Analyses-A 32 P-labeled primer/template complex was extended in the presence of varying concentrations of a single dNTP. The molar ratio of Dpo4 to DNA was Ͻ0.10. Polymerase concentrations and reaction time were adjusted to keep the extent of product formation Ͻ20% (50). All determinations were usually done at twelve dNTP concentrations (with the K m values in the center of the points). Reactions were quenched by the addition of EDTA-formamide solution, analyzed using gel electrophoresis, and then quantitated (phosphorimaging). Graphs of product formation versus dNTP concentration were fit using nonlinear regression (hyperbolic fits) using GraphPad Prism Version 3.0 (San Diego, CA) for the determination of k cat and K m values.
Pre-steady-state Reactions-Rapid quench experiments were performed using a model RQF-3 KinTek Quench Flow Apparatus (KinTek Corp., Austin, TX), with 50 mM Tris-HCl (pH 7.4, 25°C) aqueous solution in the drive syringes. Reactions were initiated by rapid mixing of 32 P-primer/template/Dpo4 mixtures (12.5 l) with the dNTP⅐Mg 2ϩ complex (10.9 l) and were then quenched with 0.6 M EDTA after varying times. 20-l aliquots of the reaction product solution were mixed with 100 l of EDTA-formamide solution, and the products were separated using denaturing gel electrophoresis (see above) and quantitated. Pre-steady-state data, with excess DNA or Dpo4, were fit to Equations 1 or 2, respectively, using nonlinear regression analysis in GraphPad Prism Version 3.0, where y is the concentration of product, A is the burst amplitude, k p is the pre-steady-state rate of nucleotide incorporation, t is time, and k ss is the steady-state rate of nucleotide incorporation (51,52).
Pre-steady-state Trap Experiments-Pre-steady-state incorporation of dCTP opposite N 2 -AnthG was also analyzed in a RQF-3 KinTek Quench Flow Apparatus, with excess unlabeled trap DNA added along with dCTP. Reactions were initiated by rapid mixing of a 70 nM Dpo4 plus 120 nM 32 P-labeled 24-mer primer/36-mer template (containing N 2 -AnthG) complex in 50 mM Tris-HCl buffer (pH 7.4) with a second solution containing 1 mM dCTP, 5 mM Mg 2ϩ , and 1.2 M unlabeled trap 24-mer/ 36-mer in the same buffer. Reactions were quenched with 0.6 M EDTA after varying times. The reaction products were quantified by denaturing gel electrophoresis and the data points were fit to Equation 1.
Phosphorothioate Analysis-Phosphorothioate analysis was done to compare the burst rates for the incorporation of dCTP and (S p )-dCTP␣S opposite the template bases. Reactions were initiated by rapid mixing of 70 nM 32 P-primer/template and 120 nM Dpo4 in 50 mM Tris-HCl buffer (pH 7.4, 12.5 l) with a mixture of 1 mM dNTP or (S p )-dNTP␣S and 5 mM MgCl 2 solution (10.9 l) in the quench flow apparatus and were then quenched with 0.6 M EDTA after varying reaction times. The reaction products were quantified by denaturing gel electrophoresis and the data points were fit to Equation 1.
Determination of k pol and K d,dCTP -k pol and K d,dCTP were estimated by performing pre-steady-state reactions in the quenched-flow apparatus with varying reaction times at different dNTP concentrations. Graphs of the burst rates (k p ) versus dNTP concentration were fit to a hyperbolic equation, where k pol is the maximal rate of nucleotide incorporation and K d,dCTP is the equilibrium dissociation constant for dCTP (51,52).

LC-MS/MS Analysis of Oligonucleotide Products from Dpo4
Reactions (46,53)-Dpo4 (5 M) was preincubated with primer/template (10 M), and the reaction was initiated by addition of dNTP (1 mM each) and MgCl 2 (5 mM) in 50 mM Tris-HCl buffer (pH 7.4) at 37°C for 4 h. Reactions were terminated by extraction of the remaining dNTPs using a size-exclusion chromatography column (Bio-Spin 6 chromatography column, Bio-Rad). Concentrated stocks of Tris-HCl, dithiothreitol, and EDTA were added to restore the concentrations to 50 mM, 5 mM, and 1 mM, respectively. E. coli uracil DNA glycosylase (20 units, Sigma-Aldrich) was added, and the solution was incubated at 37°C for 6 h to hydrolyze the uracil residue on the extended primer. The reaction mixture was then heated at 95°C for 1 h in the presence of 0.25 M piperidine, followed by removal of the solvent by centrifugation under vacuum. The dried samples were resuspended in 100 l of H 2 O for MS analysis.
LC-MS/MS analysis was performed on an Acquity UPLC system (Waters, Milford, MA) connected to a Finnigan LTQ mass spectrometer (Thermo Fisher Scientific, Waltham, MA), operating in the ESI-negative ion mode. An Acquity UPLC BEH octadecylsilane (C 18 ) column (1.7 m, 1.0 mm ϫ 100 mm) was used with the following LC conditions: Buffer A contained 10 mM CH 3 CO 2 NH 4 plus 2% CH 3 CN (v/v), and buffer B contained 10 mM CH 3 CO 2 NH 4 plus 95% CH 3 CN (v/v). The following gradient program was used, with a flow rate of 150 l min Ϫ1 : 0 -2.5 min, linear gradient from 100% A to 95%A/5% B (v/v); 2.5-6.0 min, linear gradient to 75% A/25% B (v/v); 6 -6.5 min, linear gradient to 100% B; 6.5-8.0 min, hold at 100% B; 8.0 -9.0 min, linear gradient to 100% A; 9.0 -12.0 min, hold at 100% A. The temperature of the column was maintained at 50°C. Samples were injected with an autosampler system. ESI conditions were as follows: source voltage, 4 kV; source current, 100 mA; auxiliary gas flow rate setting, 20; sweep gas flow rate setting, 5; sheath gas flow setting, 34; capillary voltage, Ϫ49 V; capillary temperature, 350°C; and tube lens voltage, Ϫ90 V. MS/MS conditions were as follows: normalized collision energy, 35%; activation Q, 0.250; and activation time, 30 ms. The doubly charged (negative ion) species were generally used for CID analysis. Calculations for the CID fragmentations of oligonucleotide sequences were done using a program linked to the Mass Spectrometry Group of Medicinal Chemistry at the University of Utah.
Kinetic Simulations-Kinetic simulations were performed using two mechanisms: a minimal mechanism, which has been previously defined for DNA polymerization (54,55), and a mechanism with the presence of a non-productive ternary complex (56,57). The fits to the experimentally determined data were developed utilizing DynaFit (BioKin, Watertown, MA) (58), run on an iMac computer using a Macintosh OS 8.5.1 operating system (Apple Computer, Cupertino, CA).
Crystallization and X-ray Diffraction Data Collection-The 18-mer template containing N 2 -NaphG and 14-mer primer containing a 3Ј-terminal dideoxycytidine ( Table 1) were annealed to form a duplex. Dpo4 was mixed with DNA (1:1.2 molar ratio) in 20 mM Tris-HCl buffer (pH 8.0, 25°C) containing 60 mM NaCl, 4% glycerol (v/v), and 5 mM ␤-mercaptoethanol and then placed on ice for 1 h prior to incubation with 5 mM MgCl 2 and 1 mM dGTP. The final Dpo4 concentration was ϳ10 mg ml Ϫ1 . Crystals were grown using the sitting drop/vapordiffusion method with the reservoir solution containing 20 mM Tris-HCl (pH 8.0 at 25°C), 15% polyethylene glycol 3350 (w/v), 60 mM NaCl, 5 mM MgCl 2 , and 4% glycerol (v/v). Droplets consisted of a 1:1 (v/v) mixture of the Dpo4⅐DNA⅐Mg 2ϩ ⅐dGTP complex and the reservoir solutions and were equilibrated against the reservoir solutions. Crystals were soaked in mother liquor containing an additional 25% polyethylene glycol 3350 (w/v) and 15% ethylene glycol (v/v) and then swiped through paratone-N (Hampton Research, Aliso Viejo, CA) and flash frozen in a stream of liquid nitrogen.
Structure Determination and Refinement-A refined wildtype Dpo4-G complex (pdb accession code 2bqr (46)) minus solvent molecules, template residue 5, metal ions, and dGTP was used as the starting model for Npg-1. Molecular replacement was performed using MOLREP as a part of the CCP4 program suite (61). The refined structure of the Npg-1 complex (minus solvent molecules, template residue 5, metal ions, and dGTP) served as the starting model for the Npg-2 structure, and the locations of the individual models were optimized by sev-eral rounds of rigid body refinement while gradually increasing the resolution of the diffraction data.
Manual model rebuilding was done with the program TUR-BO-FRODO. 4 The maps were computed using the sA-modified coefficients (62). Clear positive density for the Mg 2ϩ ions and the dGTP was observed in the initial difference Fourier electron density maps of both complexes, although the positive density was sparser for dGTP in the Npg-1 structure. Positive density was observed for the N 2 -NaphG-modified template residue, but several rounds of modeling and refinement were required to ascertain the final orientation of the base and the naphthalene moiety in both structures. The CNS package (63) was used for the refinement of the models by performing simulated annealing, gradient minimization, and refinement of individual isotropic temperature and occupancy factors. The statistics of the refined models for all structures are summarized ( Table 2). The crystallographic figures were prepared using PyMOL. 5 Calculation of the DNA helical parameter descriptions was performed using CURVES (64). Fig. 2. An increase in the Dpo4 concentration enhanced extension for all DNA adducts examined. Dpo4 readily extended beyond the unmodified G and N 2 -MeG, N 2 -EtG, N 2 -IbG, N 2 -BzG, and N 2 -NaphG templates to yield 35-, 36-, and 37-mer products, although the extensions were partially blocked opposite these adducts to yield (unelongated) 25-mer product. Other products (28-, 29-, and 31-mers) were also produced. With N 2 -AnthG, only a onebase extension product was observed, and subsequent extension was severely blocked under these conditions.

Primer Extension by Dpo4 Using All Four dNTPs-Fulllength extension beyond various N 2 -alkyl G adducts in the presence of all four dNTPs by Dpo4 is shown in
Steady-state Kinetics of dNTP Incorporation by Dpo4 Opposite G and N 2 -Alkyl G Adducts-Steady-state kinetic parameters (k cat and K m ) were measured for the single incorporation of each dNTP opposite G and N 2 -alkyl G adducts, with varying concentrations of dNTP (Table 3). Dpo4 preferentially incorporated dCTP opposite all of the N 2 -alkyl G adducts. However, the efficiencies (k cat /K m ) for correct incorporation opposite N 2 -alkyl G adducts were decreased 3-to 125-fold compared with unmodified G, due to both the decreased catalytic rates (k cat ) and increased K m,dNTP values. The efficiencies for the misincorporation of dATP, dGTP, and dTTP opposite all the N 2 -alkyl G adducts were similar to those measured with unmodified G.
Correct incorporation of dCTP opposite unmodified G was preferred over misincorporation for Dpo4, yielding a misincorporation frequency (defined as f ϭ (k cat /K m ) dNTP /(k cat / K m ) dCTP , where dNTP dCTP) of 10 Ϫ5 . The misincorporation frequency was 10 Ϫ3 to 10 Ϫ4 for all the N 2 -guanyl adducts (10to 100-fold increase), mainly due to decreased correct incorporation efficiency (as opposed to changes in the misincorporation efficiency).
Steady-state Next-base Extension following C or T Paired with G or N 2 -Alkyl G Adducts-Steady-state next-base extension (i.e. incorporation of dGTP opposite the next template base, C (Table 4)) was examined; only the extension from mismatched G (or N 2 -alkyl G):T was studied here because all misincorporation events opposite the N 2 -alkyl G adducts showed similar efficiencies. Extension efficiencies (k cat /K m ) were decreased 10to 74-fold (for extension from the N 2 -alkyl G:C pairs, compared with the unmodified G:C pair). The N 2 -alkyl G adducts showed little effect on the extension efficiency from the mispairs of N 2 -alkyl G adducts (or the G:T mismatch). Overall, extension from the matched base pair was still more efficient than from a mismatched pair for all the N 2 -alkyl G adducts; the misextension ratio (f ext ) (defined as f ext ϭ (k cat /K m ) mispair /(k cat /K m ) correct pair (50,65)) was increased 6-to 22-fold for the N 2 -alkyl G adducts (compared with unmodified G).
Pre-steady-state Burst Kinetics of dCTP Incorporation Opposite G or N 2 -Alkyl G Adducts by Dpo4-Dpo4 clearly showed burst phases for the correct incorporation of dCTP opposite unmodified G or N 2 -alkyl G adducts, even with the bulky adduct N 2 -AnthG (Fig. 3). The burst phase measures incorporation of dCTP opposite G or modified G during the initial binding event between enzyme and substrate. The linear phase of the plot results from subsequent binding events and is dependent upon a conformational relaxation from the active conformation following pyrophosphate release (66). The burst phase for dCTP incorporation opposite N 2 -alkyl G adducts showed a trend toward decreased amplitude as the adduct bulk increased from N 2 -MeG to N 2 -AnthG, and burst rates were also attenuated compared with the unmodified G.
Steady-state and Pre-steady-state Burst Kinetics of dTTP Incorporation Opposite G and N 2 -Alkyl G Adducts by Dpo4-No bursts were detected for the incorporation of dTTP opposite G or any N 2 -alkyl G adducts, except for N 2 -MeG (Fig. 4A). The incorrect incorporation of dTTP opposite N 2 -AnthG yielded a higher misincorporation rate (Fig. 4A). Misincorporation of both dTTP and (S p )-dTTP␣S opposite N 2 -MeG showed small bursts (Fig. 4B), with much lower burst amplitudes (ϳ3% of the enzyme concentration), burst rates, and steady-state rates.
Determination of k pol and K d,dCTP Values for dCTP Incorporation by Dpo4-Analysis of the changes in the pre-steady-state burst rates as a function of increasing dNTP concentration (supplemental Fig. S1) yielded maximum rates of nucleotide incorporation (k pol ) and apparent binding affinity of the dCTP to the Dpo4⅐DNA binary complex to form a ternary complex poised for catalysis (K d,dCTP ) (51, 52). The incorporation of dCTP opposite unmodified G or N 2 -MeG showed similar k pol (ϳ3.0 s Ϫ1 ) and K d,dCTP (ϳ10 M) values (Fig. 5, A and B).
where the outer sum (hkl) is taken over the unique reflections. c R f ϭ ⌺ hkl Fo hkl ͉Ϫk͉Fc hkl /⌺ hkl ͉Fo hkl ͉, where ͉Fo hkl ͉ and ͉Fc hkl ͉ are the observed and calculated structure factor amplitudes, respectively. d R free idem, for the set of reflections (5% of the total) omitted from the refinement process. e c-v, cross-validation. FEBRUARY 6, 2009 • VOLUME 284 • NUMBER 6 all the N 2 -alkyl G adducts (8 to 29 M) were similar to that of unmodified G (10 M) (3-fold variation compared with G).
Pre-steady-state Incorporation of dCTP Opposite N 2 -AnthG with Excess Trap DNA-A 10-fold molar excess of unlabeled unmodified DNA was used to trap free Dpo4 after Dpo4 dissociated from the 32 P-labeled DNA containing N 2 -AnthG; this experiment was done to establish whether a burst was indeed observed, because only Dpo4 that was still bound to the labeled DNA could extend the (radiolabeled) primer during the single binding event (24). Incorporation of dCTP opposite N 2 -AnthG with or without trap DNA showed a burst phase, and the burst amplitudes were similar (Fig. 6). A slow linear phase is observed in the trap experiment, most likely due to the relatively low burst in product formation (ϳ17% of [Dpo4]) and the fact that the protein trap is not 100% efficient. The major point of the trapping experiment was to be certain that the small burst amplitude observed with the N 2 -AnthG-containing substrate is representative of true pre-steady-state conditions.
Analysis of Primer Incorporation/Extension Products using LC-MS/MS-Full-length extension assays showed that all of the adducts, except for N 2 -AnthG, can be readily extended (Fig.  2). LC-MS/MS was utilized for the sequence analysis of the oligonucleotide extension products (46,67). A key element for the success of this method is the placement of uracil residues in the primer strand and the use of uracil DNA glycosylase to hydrolyze at the introduced uracil sites, to reduce the product length for MS sequence analysis (46). Each product in the oligonucleotide product mixtures could be readily identified and quantified by LC-MS/MS (Table 5, supplemental Figs. S3, S4,  and S5, and Tables S1-S4). From unmodified G to the

Steady-state kinetic parameters for one-base incorporation opposite G and N 2 -alkyl G adducts by Dpo4
The extent of conversion of primer to the product was kept Ͻ20% by adjustment of the enzyme concentration and reaction time.

-alkyl G):C (or T) template primer termini by Dpo4
The extent of conversion of primer to the product was kept Ͻ20% by adjustment of the enzyme concentration and reaction time. N 2 -NaphG template, all of the detected products corresponded to the correct incorporation of dCTP opposite G or the N 2 -alkyl G adduct, with accurate extension following (as well as some blunt-end addition of A or C) ( Table 5). For the bulkiest adduct, N 2 -AnthG, 79% of the products resulted from (correct) dCTP incorporation and the other 21% corresponded to misincorporation of dATP opposite N 2 -AnthG, with error-free extension following. No frameshift products were detected in any case. Crystal Structures of an N 2 -NaphG⅐C Complex-Two x-ray structures were determined for Dpo4 in complex with a DNA duplex in which the template strand contains an N 2 -NaphG residue. A 14-base primer was designed such that the dideoxycytosine (C dd ) at the terminus should pair with the N 2 -NaphG template residue. Both structures contained three Mg 2ϩ ions total (two coordinated in the active site) and an incoming dGTP, putting N 2 -NaphG in the post-insertion (Ϫ1) position. Each of the crystals diffracted to ϳ3 Å, with reasonable R merge values (9.4% and 7.8%, for Npg-1 and Npg-2, respectively). Both crystals belong to the space group P2 1 2 1 2, with a single complex per asymmetric unit.

Template Primer
The two structures are very similar overall (Fig. 7A). Analysis of the helical parameters of the DNA substrates reveals perturbations that are observed in both complexes. Similar to results obtained with O 6 -BzG-modified templates (42), each of the two refined structures containing N 2 -NaphG possesses a distinct arrangement of the DNA in the active site. The naphthyl moiety exhibits well defined electron density for both orientations (Fig.  7). A simulated annealing omit map (excluding template N 2 -NaphG, primer terminal ddC, incoming dGTP, and active site Mg 2ϩ ) further supports the correct positioning of the residues bound in the active site of both complexes (supplemental Fig. S6). The N 2 -NaphG residue is found in the syn orientation in the Npg-1 complex, which allows hydrogen bonding to occur between the exocyclic amino groups of the last two cytosines of the primer with the O6 atom of N 2 -NaphG (Fig. 7C). The ␣-phosphate of the incoming dGTP is positioned ϳ5 Å from the primer terminus, and two of the Mg 2ϩ ions are coordinated in the active site of the enzyme by residues Asp-7, Asp-105, Glu-106, and Asp-156. Overall, Npg-1 would appear to be an "active" complex.
The N 2 -NaphG residue is found in the anti orientation in the Npg-2 complex but the terminal C of the primer is largely disordered. Based on the density surrounding the terminal phosphate and sparse density in the region of the base, the dideoxycytosine was flipped into the minor groove, as has been observed in other non-productive Dpo4 structures (38,42).
The N 2 -NaphG residue is found in the anti orientation in the Npg-2 complex. The terminal dideoxycytosine of the primer is flipped into the minor groove, based on the density surrounding the terminal phosphate and sparse density in the region of the pyrimidine ring system. Such a conformation has been observed in other non-productive Dpo4 structures (38,42).   FEBRUARY 6, 2009 • VOLUME 284 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 3569
There is considerable distortion of the DNA duplex in both Npg-1 and Npg-2 structures when compared with the structure with a DNA lesion that replicated with high efficiency (dCTP opposite 8-oxoG by Dpo4; pdb ID code 2c2e; Fig. 7B). In both instances the base pairs to the 3Ј-side of the N 2 -NaphG residue are severely buckled (35°and 24°for Npg-1 and Npg-2, respectively), with large tilts for the cytosine at the Ϫ2 position (t ϭ 27°and t ϭ 19°for Npg-1 and Npg-2, respectively). Furthermore, the base pairing distance between hydrogen bond donors/acceptors is increased several residues upstream from N 2 -NaphG, indicating that the lesion can distort the helix several base pairs away from the actual site of damage.
Kinetic Simulation to a Mechanism-A general minimal mechanism is shown in Fig. 8A (51, 54, 55, 68 -72), modified with the addition of step 8 (56,66). The binding affinity and a burst rate for the incorporation of dCTP opposite G and N 2 -al-kyl G adducts by Dpo4 were initially set at 15 M and 3 s Ϫ1 , respectively, based on the pre-steady-state kinetic determinations (Fig. 5). Recently, our laboratory utilized mutant Dpo4 containing Trp as a fluorescent probe to rigorously examine the kinetics describing Dpo4-catalyzed polymerization (66). It was concluded that a slow conformational relaxation following pyrophosphate release is the major rate-limiting determinant in the Dpo4 catalytic cycle. The kinetic mechanism used in our modeling studies reflects these new insights into Dpo4 catalysis (66).
The minimal mechanism fit generally well for the DNA polymerization with a stoichiometric burst amplitude. In this study, the incorporation of dCTP opposite G and N 2 -alkyl G adducts (except N 2 -MeG) by Dpo4 resulted in reduced burst amplitudes (Fig. 5). Simulation of experimental results for partial bursts cannot yield a reasonable fit for G itself or the other

5Ј-GCCUCGAGUCAGCCGUAGACGU AG 3Ј-CGGAGCTCGGTCGGCG TCT GCG TCG CTC CTG CGG CT
N 2 -alkyl G adducts using the minimal mechanism (except for N 2 -MeG), as expected (56), even with adjustment of the rates of steps after phosphodiester bond formation (step 4 in Fig. 8A).
An alternate mechanism with an additional non-productive E-DNA-dNTP ternary complex (56, 66) (Fig. 8A) yielded good fits to the experimental results for the incorporation of dCTP opposite G (Fig. 8B) and all of the N 2 -alkyl G adducts (except for N 2 -MeG, Fig. 8C), using the experimentally measured enzyme concentration (Fig. 8, C-H). The same rate constants shown in the scheme of Fig. 8A were used in the simulations to generate values of k 4 , k 8 , and k Ϫ8 (Table 6), and the ratios for the conver-sion of intermediate to the non-productive complex versus to the product (k 8 /k 4 ) varied from 0.23 (N 2 -MeG) to 4.8 (for N 2 -AnthG). It should be noted that any product arising from non-productive complexes that escape the kinetic side reaction (k 8 /k Ϫ8 ) does not appear to greatly alter the progress curve, because all of the burst experiments are best fit by single-exponential and not double-exponential equations.

DISCUSSION
Although the Y-family DNA polymerase Dpo4 has been used extensively as a model for translesion polymerization, systematic studies on its ability to incorporate past lesions of different sizes have not been done previously. We had a set of oligonucleotides differing only in their N 2 -substitution at a single G site available and used these in this study with Dpo4. As in the cases of the human Y-family polymerases we have analyzed (24,(27)(28)(29)48), the catalytic efficiency (steady-state k cat /K m ) showed relatively limited attenuation with increasing size of the lesion at the G N2 atom, in contrast to replicative DNA polymerases (e.g. human pol ␦, HIV-1 reverse transcriptase, and bacteriophage pol T7 Ϫ ) (24). Very recently RNA polymerases have been found to vary in their abilities to insert and copy past N 2 -EtG (16). With Dpo4 and this series of N 2 -guanyl derivatives, the fidelity showed only a limited decrease until the size increased to that of N 2 -AnthG (Table 5). Of interest was the maintenance of a pre-steady-state kinetic burst throughout the entire series of N 2 -guanyl substitutions, with decreasing burst amplitude. The K d,dCTP did not change very much with increasing bulk at the guanyl N2 atom.
Two crystal structures of Dpo4 with template DNA containing N 2 -NaphG were solved. In both cases, the N 2 -NaphG adduct was positioned opposite dideoxycytidine at the 3Ј-end of a primer (at the Ϫ1 position, i.e. post-insertion), but two different active site orientations were observed. The N 2 -NaphG moiety adopts the syn orientation in the Npg-1 structure and the anti conformation in the active site of Npg-2. The DNA duplex suffers some perturbation in both structures, including a slight widening of the major groove near the primer⅐template FIGURE 7. Comparison of the Npg-1 and Npg-2 structures. A, superimposition of the refined Npg-1 (cyan) and Npg-2 (purple) structures reveals similar overall topology (r.m.s.d. 0.52). B, the DNA substrates from Npg-1 and Npg-2 were superimposed with the DNA from the ternary complex of Dpo4 inserting dCTP opposite 8-oxoG (orange, pdb ID code 2c2e). A more pronounced bending of the helical axis was observed in the N 2 -NaphGmodified DNA. Other perturbations in the N 2 -NaphG DNA, e.g. buckling of the bases and widening of the major groove near the primer⅐template junction, are also evident. C, the active site of the Npg-1 crystal structure is shown in schematic form. The 3F o Ϫ 2F c electron density map (cyan mesh) is contoured at the 1 level around the N 2 -NaphG residue. The incoming dGTP (red) is shown along with the magnesium ions (green spheres). D, base pairing and stacking interactions observed in the Npg-1 structure. E, the active site of the Npg-1 crystal structure is shown in schematic form. The 3F o Ϫ 2F c electron density map (purple mesh) is contoured at the 1 level around the N 2 -NaphG residue. The incoming dGTP (red) is shown along with the magnesium ions (green spheres). F, base pairing and stacking interactions observed in the Npg-2 structure. junction and buckling of the base pairs upstream from the adduct site. Of the two structures, the Npg-1 complex appears to be productive, whereas the Npg-2 complex is considered non-productive.
In the structure Npg-2, the anti form of N 2 -NaphG has the primer 14-C flipped out of the active site and distorted (Fig. 7F), indicating that incorporation of dCTP opposite the bulkier N 2 -guanyl adducts will be very unfavorable, with a tendency to form less productive complexes. In comparison to the Watson-Crick base pairing presumably used for dCTP with template G and small N 2 -alkyl G adducts, the fraction of non-productive complex(es) is presumably increased with N 2 -guanyl adduct bulkiness, and incorporation burst amplitudes are decreased (Fig. 3). Kinetic modeling also yielded an increase in the residence time in the non-productive complex with N 2 -adduct bulkiness ( Table 6 and Fig. 8). The bulkier N 2 -NaphG can also be in a syn configuration (Npg-1), which we conclude can incorporate dCTP opposite the Hoogsteen face of a modified G (Fig.  7). Thus the N 2 -alkyl G adducts may be incorporating dCTP via either of two (or more) mechanisms: with the small N 2 -alkyl G adducts, mainly via the usual anti configuration/Watson-Crick pairing mode; for the bulkier N 2 -aralkyl G adducts, via a syn/ Hoogsteen mode. Changes in the use of these mechanisms with increasing bulkiness of the adducts may be reflected in the variations of the observed kinetic parameters (Fig. 3).
Of the two structures solved here, the complex that contains syn N 2 -NaphG in a Hoogsteen-type pairing with ddC appears to be the more productive. However, the use of a syn-oriented template base would not be expected to yield the level of fidelity observed for Dpo4 (Table 3). The Y-family member pol has been shown to rely upon Hoogsteen base pairing even with unmodified DNA (73,74). In biochemical terms, pol has an A, general mechanism. The mechanism without step 8 and the non-productive complex is termed the "minimal mechanism" in the text.
Step 8 was used in fitting (Table 6). E, polymerase; D n , DNA substrate; E*, conformationally modified polymerase; D nϩ1 , DNA extended by one base (product); E#, inactive polymerase complex; and PPi, pyrophosphate. Forward and reverse rate constants for each step were developed on the basis of simulations of pre-steady-state nucleotide incorporation reactions with unmodified substrates. Parts B-H, the values shown in Part A and the values of k 4 , k 8 , and k Ϫ8 presented in Table 5 were used in fitting burst experiment progress curves for each of the templates (data points from Fig. 3).

TABLE 6
Kinetic modeling of pre-steady-state rate constants in a DNA polymerase mechanism Indicated rates and errors were estimated using the program Dynafit. See Fig. 8A for mechanism, Fig. 8  unusual incorporation spectrum in that it incorporates opposite template dA with much greater efficiency and fidelity than other bases (75). When incorporating opposite template dG, pol does not possess the level of fidelity observed for any of the N 2 -alkyl G adducts tested with Dpo4. Crystal structures of pol have shown that the Hoogsteen pairing mode may arise, because the template sugar is embedded in a "cavity" formed primarily by residues from the finger domain (specifically Leu-62, Lys-60, and Gln-59), which shortens the C1Ј-C1Ј distance from 10.65 Å (anti) to 8.6 Å (syn) (74). Such a shortening of the C1Ј-C1Ј distance is not observed in the Npg-1 structure, but a direct comparison cannot be made because the syn-oriented G in the pol structure is in the insertion context, while syn N 2 -NaphG is in the post-insertion context of Dpo4. Nevertheless, some differences in this region are worth noting. Of the three residues postulated to influence the C1Ј-C1Ј distance in pol , only Gln-59 is distinctive from the Val-32 of Dpo4, because the amide side chain of Gln-59 protrudes into the region of the active site that contains the template base. The Lys-60 of pol contacts the phosphate backbone of the template DNA, whereas the analogous residue in Dpo4 (Phe-33) marks the beginning of a loop between ␤-sheets 2 and 3 in the finger domain. The loop is comprised of eight amino acids residues in Dpo4, which contrast to the short three-residue ␤-hairpin observed in pol . Previous work has shown that pol is not particularly tolerant of bulk at the N2 atom of guanine and suffers a greater loss of fidelity during bypass of N 2 -dG adducts compared with either Dpo4 or pol (28,48). The fidelity observed for pol bypass of N 2 -alkyl G adducts is ϳ100to 1000-fold more error-prone than the results obtained with Dpo4. Based on the crystal structures reported here and previous kinetic analysis it appears that the molecular features resulting in Dpo4-catalyzed bypass of N 2 -alkyl G adducts are not homologous to the ones that influence the ability of pol to preferentially utilize Hoogsteen pairs. Comparison to pol suggests that perhaps we have not captured the true active complex representing Dpo4-catalyzed bypass of N 2 -NaphG. The presence of two substrate orientations in the Dpo4 active site is reminiscent of a study in which the trans N 2 -(10S) BPDE-G adduct was observed in two different orientations when bound to Dpo4 (45). Modeling studies of Dpo4-catalyzed bypass of a bulky N 2 -BPDE-G adduct had predicted that the adduct could be accommodated in both syn and anti orientations during insertion opposite the lesion (76). The resulting orientations observed in the N 2 -BPDE-G crystal structures were distinct from the models, but it should be noted here that the modeling study focused on insertion, whereas in the crystal structures the N 2 -BPDE-G adduct is in the post-insertion context. In the "productive" crystal structure (BPG-1B) (45), the N 2 -BPDE-G moiety is flipped into a gap between the finger and little finger domains, which was not predicted by molecular modeling. The BPG-1B complex of Bauer et al. (45) appears to be associated with Ϫ1 frameshift events, as judged by the available gel data (i.e. the predominant dNTP incorporation is the one corresponding to the residue 5Ј of the BPDE adduct in the template). These events would argue that the "Type II" crystal structure is most relevant to the observed function during bypass of N 2 -BPDE-G. No quantitative kinetic results were reported, so comparisons to the N 2 -guanyl adducts used here are not possible. The syn orientation of the N 2 -BPDE-G adduct in the Dpo4 active site results in a complex that is non-productive, because the BPDE ring system is stacked between the last two bases in the primer and this distortion prevents optimal binding of the incoming dATP. By way of comparison, Dpo4 places the naphthyl group between the template DNA and the little finger domain when the N 2 -NaphG moiety in the syn orientation, not into the "gap" observed for the BPDE adduct (supplemental Fig. S6). With the placement of the N 2 -NaphG moiety in the syn orientation, Dpo4 can achieve what appears to be a productive post-insertion complex. However, there is significant distortion to the base pairs adjacent to the lesion and the major groove is widened by ϳ3-4 Å starting three base pairs upstream of the adduct site. Superimposition of the Npg-1 complex with BPG-1B (45) reveals that, in contrast to BP-dG, the purine ring of N 2 -NaphG remains stacked in the template strand, perhaps due to the added flexibility of a methylene group between the exocyclic amino group and the naphthalene moiety (supplemental Fig. S6). The major similarity between the N 2 -NaphG structures solved here and the N 2 -BPDE-G structures is the ability of Dpo4 to allow several substrate conformations without greatly altering the polymerase structure.
It is of interest to compare the Npg-1 and Npg-2 structures solved here with the reported structure of pol bound to unmodified DNA (77), because Dpo4 and pol are both members of the DinB subfamily and can be considered homologues. Superimposition of the complexes reveals that, not surprisingly, the enzymes are quite similar in domain arrangement and orientation of the DNA substrate (Fig. 9A). Closer inspection of the polymerase active site shows that pol can clearly accommodate the N 2 -NaphG adduct in the syn orientation (Fig. 9B). Furthermore, the N-terminal portion of pol , which forms a "clasp" that encircles the DNA, has several aromatic side chains that could in theory promote the syn orientation of aromatic N 2 -aralkyl G adducts during translesion synthesis by way of interactions. Such a hypothesis seems somewhat counterintuitive. One might initially predict (as we did) that minor groove adducts could be housed in the larger active sites of Y-family DNA polymerases, thereby allowing normal Watson-Crick geometry to occur during bypass. A computational study of pol insertion opposite N 2 -BPDE-dG and N 6 -BPDE-dA adducts predicted that the enzyme can effectively bypass N 2 -BPDE-dG in the anti orientation (78). The same study postulates that the N-clasp is important for "near error-free" bypass of N 2 -BPDE-dG and blockage of pol by the N 6 -BPDE-dA adduct. The authors propose that the N-clasp promotes the anti orientation during accurate bypass of N 2 -BPDE-dG. Even though a productive configuration is not observed in the Npg-2 structure, we cannot rule out the possibility that accurate bypass of the larger N 2 -alkyl G adducts proceeds through some orientation that is closer to normal Watson-Crick geometry than the pairing mode observed in the Npg-1 complex. Because the N-clasp is unique to pol , it may play a role in promoting accurate bypass of minor groove lesions. Indeed, several lines of evidence are consistent with the idea that human pol is an important enzyme during transle- FEBRUARY 6, 2009 • VOLUME 284 • NUMBER 6 sion synthesis opposite bulky minor groove adducts and crosslinks, both at the insertion and extension steps (27,48,79,80).

Function of Dpo4 with N 2 -Guanyl Adducts
Although we did not obtain structures of the short alkyl derivatives (e.g. N 2 -MeG and N 2 -EtG), we would anticipate that these are probably very similar to the normal G complex, in that Watson-Crick pairing is very feasible, no misincorporations occur, and the bursts and rates for incorporation (of dCTP) were very similar to those observed with G (Tables 3 and 5 and Fig. 3). The base orientation observed for the larger adducts is more difficult to predict. Based on kinetic and LC-MS analysis, Dpo4 appears to favor accurate bypass of the bulky adducts over Ϫ1 frameshift deletions. Analysis of the crystal structures suggests the possibility that, as the bulk at the N2 position increases, accurate bypass of the N 2 -NaphG adduct occurs when the lesion adopts the syn orientation. Still, it is difficult to explain the limited decrease in Dpo4 fidelity observed with the N 2 -NaphG adduct (decreased ϳ20-fold relative to unmodified DNA) if the Hoogsteen face is indeed being presented to the incoming dNTP.
Our conclusion from the structural results with N 2 -NaphG is that this is a very distorted complex and the Npg-2 complex is catalytically inactive. We utilized a variation of the DNA polymerase catalytic scheme in which an inactive ternary complex is in equilibrium with the active complex (56). This model was used to fit the data for the burst plots from incorporation experiments with all of the N 2 -guanyl adducts (Fig. 8), with only the rate constants k 4 , k 8 , and k Ϫ8 allowed to vary. (The "trap" experiment results with N 2 -AnthG presented in Fig. 6 argue that the systems are operating under single-turnover conditions and that the course of each of these sets of assays occurs without dissociation of DNA from Dpo4.) The fits in Fig. 8 were good and support the model with the inactive complex. The partialburst results cannot be accommodated with a minimal mechanism in which there is no alternate complex (56). In the model, the inactive complex is not necessarily a single one, e.g. the structure described in Fig. 7 (E and F) but could be a set of inactive complexes. The model shows an increased partitioning of an initial ternary Dpo4⅐DNA⅐dNTP complex into the inactive form, as reflected in the ratio k 8 /k 4 ( Table 6). Although we have discussed the Npg-2 structure (Fig. 7) in the context of being inactive, we cannot rule out the possibility that limited pairing with C has some productive nature and occurs at a slow rate. However, we do not believe that this is a dominant complex that explains the preferential incorporation of dCTP (Table 3).
Exactly why the catalytic efficiency with N 2 -AnthG is higher than with N 2 -NaphG is unclear. The trend is more obvious with the pre-steady-state results than those for the steady state (Fig.  10). We attempted to crystallize N 2 -AnthG complexes but have not been unsuccessful. The structures of the N 2 -NaphG complexes are suggestive that adding an even larger adduct to the N2 atom might drive the conformational equilibrium toward the syn orientation, simply because the added bulk cannot be accommodated easily in the Dpo4 active site when in the anti position. Such an occurrence was certainly true of the N 2 -BPDE-G structures in which the BPDE moiety was either in the syn configuration or flipped out of the active site completely (45). The efficiency of incorporation opposite N 2 -AnthG is still diminished relative to the smaller adducts, but "stabilizing" the syn conformation could allow for greater efficiency during nucleotide insertion (with an additional effect being downstream perturbation of the DNA helix). Alternatively, N 2 -An-thG could be accommodated in a manner similar to N 2 -BPDE-dG by moving into the gap between domains.
In the course of this work we examined the "thio" effects for incorporation of dCTP opposite the N 2 -alkyl G derivatives (Fig.  3). Little effect was seen with G, N 2 -MeG, and N 2 -EtG, but major attenuations of the burst phases were seen with N 2 -IbG, N 2 -BzG, and N 2 -NaphG. A simplistic view of the results is that FIGURE 9. Structural comparison between Dpo4-catalyzed bypass of N 2 -NaphG and human pol . A, superimposition of the Npg-1 complex (cyan) and the human pol ternary complex (purple; pdb ID code 2oh2; r.m.s.d. ϭ 5.35). B, the active site DNA residues from the Npg-1 complex (cyan) are shown along with the superimposed pol structure (purple, space-filling). Hydrophobic and/or aromatic residues near the naphthyl moiety are highlighted for both enzymes.
the phosphodiester bond-formation step is not rate-limiting in the case of G and the two shorter alkyl derivatives (whereas step 3 of Fig. 8A is rate-limiting), and the bond formation step (step 4, Fig. 8A) occurs in large derivatives. The modeling provides some support for this view (Table 6). However, the thio effect was not seen with N 2 -AnthG, which does have a faster k pol value (Fig. 3). However, the modeling does not dictate that step 4 (of Fig. 8A) is rate-limiting. The difficulties in interpretation of the thio effect experiments have been discussed elsewhere (82), and the work in Fig. 3 should not be interpreted in the context of a simple competition between the rate constants k 3 and k 4 (Fig. 8A).
In conclusion, we have shown that the catalytic efficiency of the Dpo4 is largely preserved with increasing adduct size, using a homologous series of N 2 -guanyl derivatives (Figs. 1 and 10). The results have general similarity to those described for some mammalian Y-family DNA polymerases (27)(28)(29)48), although in this case burst kinetics persists throughout the entire series. Also, the fidelity of incorporation was highly preserved through the series. The structures of two bulky adduct complexes were determined (N 2 -NaphG, Fig. 7), the first of which we believe may represent an active complex (Npg-1). A second complex (Npg-2) is not predicted to be active, but we cannot rule out the possibility that some alternate form of the Npg-2 complex allows Watson-Crick pairing between dCTP and N 2 -NaphG. The kinetic courses of dCTP incorporation opposite all of the N 2 -guanyl adducts could be described using a minimal model with an inactive ternary complex in equilibrium with a catalytically competent form, although the caveat must be included that this inactive complex may be more than one and the mathematical representation may be more complex. These structure-function relationships add to our general knowledge of how Y-family DNA polymerases use their structural features to process a large variety of DNA lesions.