Mass Spectrometry Based Approach to Study the Kinetics of O⁶-Alkylguanine DNA Alkyltransferase Mediated Repair of O⁶-pyridyloxobutyl-2′-deoxyguanosine Adducts in DNA

Abstract

O⁶-POB-dG (O⁶-[4-oxo-4-(3-pyridyl)but-1-yl]deoxyguanosine) are promutagenic nucleobase adducts that arise from DNA alkylation by metabolically activated tobacco-specific nitrosamines such as 4-(methylnitrosamino)- 1-(3-pyridyl)-1-butanone (NNK) and N-nitrosonicotine (NNN). If not repaired, O⁶-POB-dG adducts cause mispairing during DNA replication, leading to G → A and G → T mutations. A specialized DNA repair protein, O⁶-alkylguanine-DNA-alkyltransferase (AGT), transfers the POB group from O⁶-POB-dG in DNA to a cysteine residue within the protein (Cys145), thus restoring normal guanine and preventing mutagenesis. The rates of AGT-mediated repair of O⁶-POB-dG may be affected by local DNA sequence context, potentially leading to adduct accumulation and increased mutagenesis at specific sites within the genome. In the present work, isotope dilution high performance liquid chromatography-electrospray ionization tandem mass spectrometry (HPLC-ESI⁺-MS/MS)-based methodology was developed to investigate the influence of DNA sequence on the kinetics of AGT-mediated repair of O⁶-POB-dG adducts. In our approach, synthetic DNA duplexes containing O⁶-POB-dG at a specified site are incubated with recombinant human AGT protein for defined periods of time. Following spiking with D₄-O⁶-POB-dG internal standard and mild acid hydrolysis to release O⁶-POB-guanine (O⁶-POB-G) and D₄-O⁶-POB-guanine (D₄-O⁶-POB-G), samples are purified by solid phase extraction (SPE), and O⁶-POB-G adducts remaining in DNA are quantified by capillary HPLC-ESI⁺-MS/MS. The new method was validated by analyzing mixtures containing known amounts of O⁶-POB-dG-containig DNA and the corresponding unmodified DNA duplexes and by examining the kinetics of alkyl transfer in the presence of increasing amounts of AGT protein. The disappearance of O⁶-POB-dG from DNA was accompanied by pyridyloxobutylation of AGT Cys-145 as determined by HPLC-ESI⁺-MS/MS of tryptic peptides. The applicability of the new approach was shown by determining the second order kinetics of AGT-mediated repair of O⁶-POB-dG adducts placed within a DNA duplex representing modified rat H-ras sequence (5′-AATAGTATCT[O⁶-POB-G]GAGCC-3′) opposite either C or T. Faster rates of alkyl transfer were observed when O⁶-POB-dG was paired with T rather than C (k = 1.74 × 10⁶ M⁻¹s⁻¹ vs 1.17 × 10⁶ M⁻¹s⁻¹).

Introduction

4-(Methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK)^* and N-nitrosonicotine (NNN) are carcinogenic nitrosamines found only in tobacco products and known to specifically induce lung cancer in laboratory animals.^1-3 The carcinogenic effects of NNK and NNN are mediated by the formation of promutagenic DNA adducts. Both tobacco-specific nitrosamines undergo cytochrome P450 monooxygenase-mediated bioactivation to form pyridyloxobutyl diazonium ions which react with nucleophilic sites in DNA. The resulting bulky pyridyloxobutylated nucleobase lesions include N7-[4-oxo-4-(3-pyridyl)-but-1-yl]deoxyguanosine (N7-POB-dG), O⁶-[4-oxo-4-(3-pyridyl)but-1-yl]deoxyguanosine (O⁶-POB-dG), O²-[4-(3-pyridyl)-4-oxobut-1-yl]thymidine (O²-POB-T), and O²-[4-(3-pyridyl)-4-oxobut-1-yl]-deoxycytidine (O²-POB-dC).^3,4 Among these, O⁶-POB-dG have been shown to be strongly mutagenic. During replication, DNA polymerases incorporate either thymine or adenine opposite O⁶-POB-dG in lieu of the correct base (cytosine), leading to G → A transitions and G → T transversions.⁵

In order to prevent mutagenesis, O⁶-POB-dG lesions must be efficiently repaired prior to DNA replication. A specialized DNA repair protein, O⁶-alkylguanine-DNA alkyltransferase (AGT), directly removes the O⁶ -alkyl group from modified guanine such as O⁶-POB-G, restoring the native DNA structure.^6-8 Upon AGT binding to double-stranded DNA, the O⁶-alkylguanosine residue is flipped out of the DNA base stack to enter the active site of the AGT protein.⁹ Histidine-146 residue in the AGT active site, with the aid of a water molecule, acts as a general base to abstract a proton from the thiol group of a neighboring cysteine-145 (Cys145).^6,9 The resulting thiolate anion accepts the alkyl group from O⁶-alkylguanosine, restoring normal guanine and forming a thioether at Cys145 (Scheme 1). This inactivates the AGT protein by inducing conformational changes, ubiquitination, and proteosomal degradation.^10,11 As a result, the AGT repair reaction is stoichiometric, inactivating one molecule of AGT protein per each alkylguanine lesion repaired.

Scheme 1.

AGT repair of O⁶-pyridyloxobutyl-dG adducts in DNA

All available evidence suggests that alkylated AGT molecules are not re-activated but are rapidly degraded by the proteasome. Several hypotheses have been put forward to explain this unusual mechanism,^8,12 which may relate to the difficulty of distinguishing O⁶-methylguanine from unmodified guanine at a high enough level of discrimination to prevent damage to normal DNA.⁸ Furthermore, although one molecule of protein is used up per adduct repaired, only a single protein is needed as compared to over 20 proteins in the nucleotide excision repair mechanism. Finally, the alkylated protein must be degraded quickly because it retains its DNA-binding abilities and may interfere with repair of additional O⁶-alkylguanine lesions.

While several studies have reported that the rates of AGT-mediated repair of O⁶-POB-dG lesions are affected by neighboring DNA bases, only a few selected DNA sequences have been examined.^13,14 Coulter et. al. used a radioflow HPLC-based methodology to show that the first order rate for AGT-mediated repair of O⁶-POB-dG placed in the first position of the H-ras codon 12 (5′-CTXGA-3′ context, 0.95 × 10⁻⁴ s⁻¹) was ~ 6 times larger for the adduct placed at the second position of codon 12 (5′TGXAG-3′ context, 0.16 × 10⁻⁴ s⁻¹).¹⁴ A much faster repair rate (0.022 s⁻¹) was observed for another sequence containing O⁶-POB-dG (X) in 5′-ATXGC-4′ sequence context.¹⁴ Similar results were reported by Mijal et. al.¹³ who employed denaturing PAGE to examine the kinetics of AGT-mediated repair of O⁶-POB-dG adducts relative to O⁶-Me-dG within rat H-ras codon 12. These authors¹³ also noted that the rate of O⁶-POB-dG repair increased when a thymine rather than cytosine was placed in the complementary strand opposite O⁶-alkyl-dG. However, to our knowledge, no systematic studies have been conducted to examine the effects of local DNA sequence on AGT-mediated repair of O⁶-POB-dG, probably due to practical limitations of the available experimental methodologies. Previous studies have relied on the ability of HPLC or gel electrophoresis methodologies to resolve alkylated and dealkylated DNA strands.^13,14 Such separations can be quite challenging, especially for longer DNA duplexes (> 20 nucleotides).

In the present work, a novel, stable isotope dilution capillary HPLC electrospray ionization tandem mass spectrometry (ESI⁺-MS/MS) methodology was developed to analyze the kinetics of AGT repair of O⁶-POB-dG adducts in vitro as a function of DNA sequence context. An application of this new methodology was demonstrated for O⁶-POB-dG lesions introduced into synthetic oligonucleotide duplexes representing p53 codon 158 and a modified rat H-ras gene sequence containing codon 12.

Experimental Procedures

Materials

Dithiane protected O⁶-POB-dG (O⁶-[3-[2-(3-pyridyl)-1,3-dithyan-2-yl]propyl]-deoxyguanosine) was synthesized according to methods published by Wang and coworkers¹⁵ and converted to 5′-dimethoxytrityl-N,N-dimethyl-formamidine-O⁶-[4-oxo-4-(3-pyridyl)but-1-yl]-2′-deoxyguanosine, 3′-[(2-cyanoethyl)-(N,N-diisopropyl)]-phosphoramidite by standard phosphoramidite chemistry.^15,16 All other nucleoside phosphoramidites, solvents, and solid supports required for the solid phase synthesis of DNA were procured from Glen Research Corporation (Sterling, VA). Human recombinant AGT protein with a C-terminal histidine tail (WT ch-AGT) was expressed in E. coli and isolated as reported previously.^17,18 The catalytic activity of the AGT protein was determined by incubating with known amounts of DNA duplexes containing O⁶-MeG.¹⁹ Approximately 75% of the total protein was active, depending on the aliquot. D₄-O⁶-POB-dG was a gift from Professor Stephen Hecht (University of Minnesota). O⁶-POB-dG-containing oligodeoxynucleotide representing a modified rat H-ras gene sequence was generously provided by Professor Lisa Peterson (University of Minnesota). Phosphodiesterase I, phosphodiesterase II, and DNase I were obtained from Worthington Biochemical Corporation (Lakewood, NJ). Trypsin was purchased from Promega (Madison, WI), and bovine intestinal alkaline phosphatase was procured from Sigma Aldrich Chemical Company (Milwaukee, WI). The rest of the chemicals were purchased either from Fisher Scientific (Fairlawn, NJ) or Sigma-Aldrich (Milwaukee, WI).

Solid Phase Synthesis of DNA

Synthetic oligodeoxynucleotides containing O⁶-[3-[2-(3-pyridyl)-1,3-dithian-2-yl]propyl]-deoxyguanosine in the context of p53 codon 158 were prepared by solid phase synthesis on a DNA synthesizer¹⁶ starting with synthetic 5′-dimethoxytrityl-N,N-dimethyl-formamidine-O⁶-[4-oxo-4-(3-pyridyl)but-1-yl]-2′-deoxyguanosine, 3′-[(2-cyanoethyl)-(N,N-diisopropyl)]-phosphoramidite.¹⁵ Following HPLC purification of synthetic oligonucleotides, the 1,3-dithiane protective group was removed with N-chlorosuccinimide to produce the corresponding O⁶-POB-G-containing DNA.^15,20

Synthetic DNA oligomers were purified by reversed phase HPLC using an Agilent 1100 HPLC system interfaced with a DAD UV detector.^21,22 HPLC fractions corresponding to oligodeoxynucleotides of interest were collected and concentrated under reduced pressure. The presence of O⁶-POB-dG was confirmed by capillary HPLC-ESI⁻ MS (Table 1).^22-24 Oligonucleotide concentrations were determined from the amounts of 2′-deoxyguanosine present in enzymatic digests using a previously published protocol.^19,25,26

Table 1.

Nucleobase sequence and molecular weight information (from HPLC2ESI²2MS) for synthetic DNA oligomers employed in this work.

Oligomer	Sequence	Molecular Weight
		Calculated	Observed
(+) p53 codon 158-O6pobG	ACCCGCGTCC[O6-POB-G]CGCCATGGCC	6026.0	6026.9
(−) p53 codon 158	GGCCATGGC GCG GACGCGGGT	6529.3	6529.4
(+) p53 codon 158	ACCCGCGTCCGCGCCATGGCC	6329.1	6328.8
(+) H-ras POB	AATAGTATCT[O6-POB-G]GAGCC	5052.3	5051.7
(−) H-ras base paired dC	GGCTCCAGATACTATT	4856.2	4855.6
(−) H-ras base paired dT	GGCTCTAGATACTATT	4871.2	4870.7

O⁶-POB-dG containing DNA duplexes were prepared by mixing equal amounts of the complimentary strands in 10 mM Tris-HCl (pH 8) buffer containing 50 mM NaCl to achieve a final concentration of 200 μM, followed by heating at 90 °C for 5 minutes and slow cooling to room temperature.²⁵

AGT reactions with O⁶-POB-dG containing DNA Duplexes

DNA duplexes containing O⁶-POB-dG (500 fmol, Table 1) were mixed with human recombinant AGT protein (400 fmol) in a final volume of 90 μl of 50 mM Tris-HCl buffer (pH 7.8) containing 0.1 mM EDTA, 0.5 mg/mL BSA, and 0.5 mM DTT. Following incubation at room temperature for fixed periods of time (0-50 seconds), repair reactions were manually quenched with 0.2 N HCl. D₄-O⁶-POB-dG internal standard (250 fmol) was added, and the DNA was subjected to mild acid hydrolysis (0.1 N HCl, 70 °C for 1 h). Samples were neutralized with ammonium hydroxide. O⁶-POB-G adducts were isolated by solid phase extraction (SPE) and quantified by the capillary HPLC-ESI⁺-MS/MS as described below.

Solid Phase Extraction

Solid phase extraction (SPE) was conducted with Strata-X cartridges (Phenomenex, Torrance, CA), which were pre-equilibrated with methanol and water. Following loading of DNA hydrolysates, cartridges were washed with 20% methanol in water. O⁶-POB-G and D₄-O⁶-POB-G were eluted with 100% MeOH. Samples were dried under reduced pressure, re-suspended in 15mM NH₄OAc (pH 5.5) containing 10% acetonitrile, and analyzed by the capillary HPLC-ESI⁺-MS/MS as described below.

HPLC-ESI⁺-MS/MS Analysis of O⁶-POB-G Adducts

Capillary HPLC-ESI⁺-MS/MS analyses were conducted using a Thermo-Finnigan Ultra TSQ mass spectrometer interfaced with a Waters Acquity capillary HPLC system. The mass spectrometer was operated in the positive ion mode. Typical spray voltage was 3.3 kV, and a capillary temperature was 250 °C. Typical tube lens offset was 91 V, and nitrogen was used as a sheath gas (43 counts). Quantitative analyses were conducted in the SRM mode, with collision energy of 13 V. Argon was used as a collision gas with a pressure of 1 mTorr. MS/MS analyses were performed with a scan width of 0.4 m/z and a scan time of 0.25 seconds.

Capillary HPLC was conducted using a Zorbax SB C18 column (0.5 mm × 150 mm, 5 μm, Agilent Technologies) eluted at a flow rate of 15 μL/min and maintained at 25°C. HPLC solvents were 15 mM ammonium acetate (A) and acetonitrile containing 15 mM ammonium acetate (10%, B). A linear gradient of 11% to 34% B in 16 minutes was used, followed by isocratic elution at 34% B for 4 minutes and column re-equilibriation.²⁶ Under these conditions, both O⁶-POB-G and its internal standard (D₄-O⁶-POB-G) eluted at ~ 14 minutes. Selected reaction monitoring (SRM) was conducted by monitoring the transitions: m/z 299.1 [M + H⁺] →148.1 [POB]⁺, m/z 299.1 →152.1 [Gua + H⁺] for O⁶-POB-G and m/z 303.1 [M + H⁺] → m/z 152.1 ([D₄-POB]⁺ and [Gua + H⁺]) for D₄-O⁶-POB-G.

HPLC-ESI⁺-MS/MS method validation

Synthetic DNA 21-mers (5′-ACCCGCGTC CGCGCCATGGCC-3′, 1 pmol) were spiked with increasing amounts of the corresponding O⁶-POB-dG-containing duplexes (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′, 0.1-1pmol). DNA mixtures were dissolved in 90 μl of 50 mM Tris-HCl buffer (pH 7.8) containing 0.5 mg/mL BSA, 0.5 mM DTT, and 0.1 mM EDTA. Following the addition of D₄-O⁶-POB-dG internal standard (250 fmol) and inactivated AGT protein (400 fmol), samples were subjected to mild acid hydrolysis (0.1 N HCl, 70 °C, 1h). The hydrolysates were neutralized with ammonium hydroxide and purified by solid phase extraction on Strata-X cartridges as described above. O⁶-POB-G and D₄-O⁶-POB-G were eluted with methanol, dried under vacuum, and re-suspended in 15 mM NH4OAc (pH 5.5) containing 10% ACN for capillary HPLC-ESI⁺-MS/MS analyses. The validation results were expressed as the ratio of HPLC-ESI⁺-MS/MS peak areas corresponding to O⁶-POB-G and D₄-O⁶-POB-G versus the actual O⁶-POB-G/D₄-O⁶-POB-G molar ratio. All analyses were performed in quadruplet.

Kinetics of POB Group Transfer in the Presence of Increasing Amounts of AGT

O⁶-POB G-containing DNA duplexes (5′-ACCCGCGTCC[O⁶-POB G]CGCCATGGCC-3′ and its compliment, 0.5 pmol, in triplicate) were dissolved in Tris-HCl buffer (50 mM, pH 7.8) containing 0.5 mg/mL BSA, 0.5 mM DTT, and 0.1 mM EDTA (final volume 90 μl). Following addition of human recombinant AGT protein (50, 200, 400, or 800 fmol), alkyl transfer reactions were allowed to proceed for varying lengths of time (0-600 seconds). Each reaction was terminated by the addition of HCl to achieve the final concentration of 0.1N. After spiking with D₄-O⁶-POB-dG internal standard (250 fmol), the samples were subjected to mild acid hydrolysis(70 °C, 1h), neutralized with NH₄OH, and purified by SPE. O⁶-POB-G adducts remaining in DNA after repair reaction were quantified by HPLC-ESI⁺-MS/MS as described above.

Second Order Kinetics for AGT Reactions with O⁶-POB-dG-Containing DNA

DNA duplexes containing site-specific O⁶-POB-G adducts (500 fmol, Table 1) were mixed with human recombinant AGT protein (400 fmol) in 50 mM Tris-HCl buffer (pH 7.8) buffer containing 0.1 mM EDTA, 0.5 mg/mL BSA, and 0.5 mM DTT (final volume, 90 μl). Samples were incubated at room temperature for 0-50 seconds and quenched with acid (final concentration, 0.1N HCl). Following the addition of D₄-O⁶-POB-dG internal standard (250 fmol), DNA was subjected to mild acid hydrolysis (1 h at 70 °C), and the unrepaired O⁶-POB-dG adducts remaining in DNA were quantified by capillary HPLC-ESI⁺-MS/MS as described above.

Second order rates (k) for AGT repair of O⁶-POB-G were calculated by plotting O⁶-POB-G concentrations repaired at time t (average of 4 separate measurements) versus time and fitting the data to the 2nd order kinetic equation (Equation 1) using the KaleidaGraph software:²⁵

$$\mathrm{k}\mathrm{t}=\frac{1}{{\mathrm{B}}_0-{\mathrm{A}}_0}\times \ln \frac{{\mathrm{A}}_0({\mathrm{B}}_0-{\mathrm{C}}_{\mathrm{t}})}{{\mathrm{B}}_0({\mathrm{A}}_0-{\mathrm{C}}_{\mathrm{t}})}$$ (1)

where A₀ is the concentration of AGT protein, B₀ is the starting concentration of O⁶-POB-dG containing DNA, and C_t is the concentration of O⁶-POB-G repaired at time t. The error in the reaction rate measurements was obtained from KaleidaGraph software by approximating how closely the experimentally obtained data points fit the defined 2^nd order kinetic equation.

Kinetics of POB Group Transfer to the AGT Protein

DNA duplexes (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′ (+) strand, 20 pmol) were dissolved in 50 mM Tris-HCl buffer (pH 7.8) containing 0.1 mM EDTA and 0.5 mM DTT. Following the addition of human recombinant AGT protein (16 pmol), the reaction mixtures (final volume, 90 μl) were incubated at room temperature for fixed periods of time (0 - 40 s). Each reaction was manually quenched by the addition of 0.5 N NaOH (90 μl). Samples were neutralized with HCl, dried under vacuum, and reconstituted in 100 mM ammonium bicarbonate buffer (pH 7.9, 50 μl). The protein was digested with trypsin (0.3 μg) for 18 hours at 37 °C.²⁷ The resulting tryptic peptides were desalted using Millipore C18 Zip Tips (Bedford, MA) and dried under reduced pressure. Samples were re-suspended in 20 μl of 0.1% formic acid and analyzed by capillary HPLC-ESI⁺-MS/MS as described below.

Capillary HPLC-ESI⁺-MS/MS analyses of tryptic peptides were performed with a Thermo-Finnigan Ultra TSQ mass spectrometer interfaced with a Waters Acquity capillary HPLC system. HPLC separation was achieved with a Zorbax 300 SB-C8 column (0.3 mm × 150 mm, 3.5 μm, Agilent Technologies) eluted at a flow rate of 6 μL/min and maintained at a temperature of 50 °C. HPLC solvents were aqueous 0.5% formic acid containing 0.1% trifluoroacetic acid (A) and acetonitrile containing 0.5% formic acid and 0.1% trifluoroacetic acid (B). A linear gradient of 3% to 55% B in 15 minutes was followed by isocratic elution at 55% B for 10 minutes. Under these conditions, unmodified AGT peptide eluted at ~ 24.1 minutes, while pyridyloxobutylated peptide eluted within ~ 24.5 min. The mass spectrometer was operated in the positive ion mode with a spray voltage of 3.5 kV, a capillary temperature of 250 °C, and tube lens offset of 103 V. Nitrogen was used as a sheath gas (25 counts).

Quantitative analyses were conducted in the MS/MS mode with collision energy of 24 V. Argon was used as a collision gas at a pressure of 1.3 m Torr, and all analyses were performed with a scan width of 0.3 m/z and a scan time of 0.15 s. Unmodified AGT active site peptide (G¹³⁶NPVPILIPCHR¹⁴⁷) was detected in the SRM mode by monitoring the transition m/z 658.4 [M + 2H]⁺² → m/z 948.6 [y₈], while the corresponding pyridyloxobutylated peptide (G¹³⁶NPVPILIP[C-POB]HR¹⁴⁷) was detected using the MS/MS transition m/z 731.8 → m/z 1095.6 [y₈] (Figure 6A, Supplement S-1). The percent pyridyloxobutylation of AGT Cys-145 residue was calculated from the HPLC-ESI⁺-MS/MS peak corresponding to the unmodified peptide containing the active site cysteine (A_unmodified) and the pyridyloxobutylated peptide (A_POB) according to Equation 2:

$$\%\text{pyridyloxobutylation of}\mathrm{Cys}145\text{of}\mathrm{AGT}=\frac{{\mathrm{A}}_{\mathrm{POB}}}{{\mathrm{A}}_{\text{unmodified}}+{\mathrm{A}}_{\mathrm{POB}}}\times 100\%$$ (2)

and plotted against reaction time (t) (Figure 6B).

Figure 6.

Formation of pyridyloxobutylated AGT protein following incubation with O⁶-POB-dG-containing DNA duplex (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′ and its compliment): Capillary HPLC-ESI⁺-MS/MS analysis of pyridyloxobutylated AGT active site peptide (G¹³⁶NPVPILIPCHR) (A). Time course for the formation of pyridyloxobutylated AGT protein (B). The line was generated by fitting the data to a single exponential equation using KaleidaGraph software.

Results

Stable Isotope labeling HPLC ESI⁺-MS/MS Approach

As discussed above, previous methodologies have relied on radioflow HPLC or gel electrophoresis of ³²P endlabeled DNA oligomers to analyze the kinetics of O⁶-POB-dG repair by AGT.^13,14 In the present work, a non-radioactive stable isotope labeling HPLC ESI⁺-MS/MS approach was developed to analyze the influence of DNA sequence context on the kinetics of O⁶-POB-G adduct repair by AGT. In our approach, synthetic DNA duplexes containing site specific O⁶-POB-dG adducts are incubated with human recombinant AGT protein (WT h-AGT) for specified periods of time, and the reactions are quenched with HCl (Figure 1).¹⁹ DNA is subjected to mild acid hydrolysis to release purines, including any O⁶-POB-G adducts remaining after the repair reaction, which are then quantified by isotope dilution - capillary HPLC electrospray ionization tandem mass spectrometry. Our initial studies (not shown) confirmed that no spontaneous removal of the POB group is observed under these conditions (0.1 N HCl, 70 °C, 1h). Quantitation is conducted in the selected reaction monitoring mode using the HPLC-ESI⁺-MS/MS peak areas corresponding to O⁶-POB-G and the deuterated internal standard (D₄-O⁶-POB-G).

Figure 1.

HPLC-ESI⁺-MS/MS methodology employed for quantitative analyses of O⁶-POB-G adducts remaining in DNA following AGT repair.

ESI⁺-MS/MS spectrum of O⁶-POB-G (m/z 229.0, [M + H]⁺) is characterized by two major fragment ions at m/z 148.1 [POB⁺] and m/z 152.1 [Gua + H]⁺ (Figure 2A). ESI⁺-MS/MS spectrum of D₄-O⁶-POB-G (m/z 303.1 [M + H]⁺) contains one main prominent peak at m/z 152.1, corresponding to both [D₄-POB⁺] and [Gua + H]⁺ (Figure 2B). Our quantitatifve method for O⁶-POB-G is based on selected reaction monitoring (SRM) of the transitions m/z 299.1 [M + H]⁺ → 148.1 [POB⁺], m/z 299.1 → [M + H]⁺ → 152.1 [Gua + H] ⁺for O⁶-POB-G and m/z 303.1 [M + H] ⁺ → 152.1 [D₄-POB⁺], m/z 303.1 [M + H] ⁺ → 152.1 [Gua + H] ⁺ for D₄-O⁶-POB-G internal standard (Figure 3).

Figure 2.

ESI⁺-MS/MS spectra of O⁶-POB-G (A) and D₄-O⁶-POB-G (B) and proposed structures of the major fragments.

Figure 3.

Capillary HPLC-ESI⁺-MS/MS analysis of O⁶-POB-G adducts remaining in DNA following AGT repair reaction. 500 fmol of double-stranded DNA (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′ and its compliment) was incubated with 400 fmol of AGT protein for 20 seconds, followed by quenching with HCl and spiking with 250 fmol of D₄-O⁶-POB-dG. The resulting mixture was subjected to acid hydrolysis and analyzed by capillary HPLC-ESI⁺-MS/MS. The mass spectrometer was operated in SRM mode by monitoring the transitions m/z 299.09 [M + H⁺] →148.1 [POB⁺], 152.07 [Gua + H⁺] for O⁶-POB-G (A) and m/z 303.09 [M + H⁺]→152.07[D₄-POB⁺], [Gua + H⁺] for D₄-O⁶-POB-G (B).

HPLC separation is achieved with a capillary Zorbax SB-C18 column (Agilent Technologies) eluted at a flow rate of 15 μl/min with a gradient of 15 mM ammonium acetate and acetonitrile. Under our conditions, O⁶-POB-G standard and D₄-O⁶-POB-G internal standard co-elute at ~ 14 min (Figure 3). Quantitative analyses were conducted by comparing the HPLC ESI⁺-MS/MS peak areas corresponding to O⁶-POB-G and D₄-O⁶-POB-G internal standard. The lower limit of detection for O⁶-POB-G was estimated as 3 fmol, with a signal to noise ratio (S/N) of 3. The lower limit of quantification of O⁶-POB-G adducts by our HPLC ESI⁺-MS/MS method was estimated as 10 fmol with a signal to noise ratio of 5.

The new HPLC ESI⁺-MS/MS method was validated by performing the analyte recovery assay. Synthetic DNA duplex (5′-ACCCGCGTCCGCGCCA TGGCC-3′ and its compliment, 1pmol) was spiked with increasing amounts of the corresponding O⁶-POB-dG-containing duplex (5′-ACCCGCGTCC[O⁶-POB-G]CGCCA TGGCC-3′ and its compliment, 0.1-1pmol) and acid-inactivated recombinant AGT protein (WT h-AGT, 400 fmol). The resulting mixtures were subjected to mild acid hydrolysis, followed by sample processing and capillary HPLC-ESI⁺-MS/MS analysis as described above. The observed O⁶-POB-G/ D₄-O⁶-POB-G ratios were plotted against the theoretical ratios of O⁶-POB-G/ D₄-O⁶-POB-G (Figure 4). A linear correlation with an R² value of 0.9965 and a slope of 1.0254 was obtained, validating our quantitative HPLC-ESI⁺-MS/MS method for O⁶-POB-G adducts in DNA.

Figure 4.

Validation results for quantitative capillary HPLC-ESI⁺-MS/MS method for O⁶-pobG using isotope dilution with D₄-O⁶-POB-G internal standard. Synthetic O⁶-pob-dG-containing DNA duplex (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′ and its compliment, 0.1-1pmol) was mixed with the corresponding native DNA duplex (1 pmol), D₄-O⁶-POB-dG internal standard, and acid-inactivated AGT protein (500 fmol)(N=4). The resulting mixtures were subjected to mild acid hydrolysis, and O⁶-POB-G adducts were quantified by capillary HPLC-ESI⁺-MS/MS. The results are expressed as a ratio of HPLC-ESI-MS/MS peak areas corresponding to O⁶-POB-G and D₄-O⁶-POB-G versus the actual O⁶-POB-G|D₄-O⁶-POB-G molar ratio.

Time Course of AGT Repair within p53 codon 158 Derived DNA Sequences

To test the applicability of our analytical methodology to studies of AGT repair kinetics, time dependent repair of O⁶-POB-dG in the presence of increasing amounts of human recombinant AGT protein was investigated. Synthetic oligodeoxynucleotide duplexes representing codon 158 of the p53 tumor suppressor gene and surrounding sequence were prepared containing O⁶-POB-dG adduct at the second position of p53 codon 158 (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′, (+) strand). Following annealing to the complementary strand, the resulting site-specifically modified duplexes were incubated with 0.1-1.6 equivalents of human recombinant AGT protein over fixed periods of time (0-600 s), and O⁶-POB-dG adducts remaining in DNA were quantified by HPLC ESI⁺-MS/MS.

The repair ratios for the control samples (R₀) consisted of peak area of O⁶-POB-G/ D₄-O⁶-POB-G in DNA alone (time t = 0 sec), while repair ratios at time = t (R_t) was obtained by analyzing ratios of area of O⁶-POB-G/ D₄-O⁶-POB-G upon incubation of DNA with AGT protein for a given time (Equation 3). The percent repair (% repair) of O⁶-POB-dG was obtained by calculating the difference in repair of control samples and repaired DNA at a given time followed by dividing this value by the repair ratio of the control sample times 100% (Equation 3).²⁵

$$\%\text{Repair}=\frac{{\mathrm{R}}_0-{\mathrm{R}}_{\mathrm{t}}}{{\mathrm{R}}_0}\times 100\%$$ (3)

A plot of % repair versus time revealed a time- and concentration-dependent removal of O⁶-POB-dG adducts by AGT (Figure 5). As expected for second order kinetics, increased AGT concentrations resulted in a proportional increase in the rate of O⁶-POB-G repair. The extent of DNA dealkylation increased proportionately until ~ 60 s, after which no additional repair was observed for the remainder of the incubation due to the depletion of active AGT protein as a result of alkyl group transfer to Cys-145 (Scheme 1). We have previously observed similar results for AGT-mediated repair of O⁶-methyl-dG adducts.¹⁹

Figure 5.

Time course for the repair of O⁶-POB-G adducts incorporated into synthetic DNA duplexes (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′ and its compliment) in the presence of increasing amounts of recombinant human AGT protein (20, 200, 400, or 800 fmol). O⁶-POB-G adducts remaining in DNA following the specified reaction times were quantified by isotope dilution HPLC-ESI⁺-MS/MS. The line was generated by fitting to a single-exponential equation using KaleidaGraph software.

Mass Spectrometry-based Detection of Pyridyloxobutylated AGT Protein

As discussed above, AGT repair reaction involves alkyl group transfer from DNA to an active site cysteine of the protein (Cys-145 in human AGT).⁷ To investigate the kinetics of protein pyridyloxobutylation in the presence of O⁶-POB-dG-containing DNA, AGT repair reactions with p53 codon 158 containing DNA duplexes (5′-ACCCGCGTCC[O⁶-POB-G]CGCCATGGCC-3′+ complement) were quenched with NaOH, and the time-dependent formation of alkylated protein was followed by capillary HPLC-ESI-MS/MS analysis of the tryptic peptide G¹³⁶NPVPILIPC¹⁴⁵HR¹⁴⁷ corresponding to AGT active site containing cysteine 145 (Cys-145). Both intact (M = 1314.4 Da) and pyridyloxobutylated AGT active site peptide (M = 1461.6 Da, POB group = 147 Da) were detected using selected reaction monitoring of the y₈ fragment formation from doubly charged pseudo-molecular ions of the peptides of interest (m/z 658.2→ 948.6 for unmodified peptide and m/z 731.8 →1095.6 for pyridyloxobutylated peptide; see Figure 6A). HPLC-ESI-MS/MS peak areas corresponding to the intact and alkylated peptide were compared to calculate the extent of protein pyridyloxobutylation. A time-dependent increase in the concentration of pyridyloxobutylated protein was observed until ~ 15 seconds, after which the concentrations of pyridyloxobutylated protein formed leveled off due to the depletion of active protein (Figure 6B). The reaction rate was faster than that in the experiments shown in Figure 5 due to the increased concentrations of DNA substrate and AGT protein used in this experiment. Higher protein amounts were required to facilitate the detection of pyridyloxobutylated active site peptide.

Kinetics of AGT Repair of O⁶-POB-G Adducts within DNA Sequence Representing codon 12 of the rat H-ras Gene

The new HPLC-ESI-MS/MS methodology was employed to analyze the kinetics of AGT-mediated POB transfer from O⁶-POB-dG-containing DNA duplexes derived from rat H-ras gene as a function of the partner base in the opposite strand. Synthetic DNA duplexes (5′-AATAGTATCT[O⁶-POB-G]GAGCC-3′, where O⁶-POB-G is placed at the first position of H-ras codon 12 were prepared containing either C or T opposite O⁶-POB-G. Following incubation of adducted duplexes with recombinant human AGT protein for defined periods of time, the reactions were quenched with acid, spiked with D₄-O⁶-POB-dG internal standard, and the amounts of O⁶-POB-G adducts remaining in DNA were determined by isotope dilution capillary HPLC ESI⁺-MS/MS. This was used to calculate the amounts of O⁶-POB-G adducts removed at time t (Figure 7). A graph of O⁶-POB-G concentrations repaired versus time was obtained, and the resulting curves were fitted to the 2^nd order kinetics equation (Equation 4) using KaleidaGraph software (Synergy Software, Reading, PA)^25,28:

$$\mathrm{k}\mathrm{t}=\frac{1}{{\mathrm{B}}_0-{\mathrm{A}}_0}\times \ln \frac{{\mathrm{A}}_0({\mathrm{B}}_0-{\mathrm{C}}_{\mathrm{t}})}{{\mathrm{B}}_0({\mathrm{A}}_0-{\mathrm{C}}_{\mathrm{t})}}$$ (4)

where A₀ is the molar concentration of AGT protein, B₀ is the molar concentration of O⁶-POB-G containing DNA duplex, C_t is the concentration of O⁶-POB-G adducts repaired at time t, t is the time in seconds, and k is the 2^nd order rate for O⁶-POB-G repair.

Figure 7.

AGT repair kinetics for O⁶-POB-dG adducts placed opposite dC or dT in synthetic DNA duplexes (5′-AATAGTATCT[O⁶-POB-G]GAGCC-3′ + strand) containing either C or T opposite the adducts). DNA (500 fmol) was incubated with human recombinant AGT protein (400 fmol) for increasing lengths of time (0-50 seconds). The reactions were stopped by HCl quenching, and unrepaired O⁶-POB-G adducts were quantified by isotope dilution HPLC-ESI⁺-MS/MS. Each data point represents an average of four independent measurements. The curves represent the best fit to a second-order exponential equation that provides the rate of O⁶-POB-dG repair by AGT obtained by the KaleidaGraph software. The error is estimated by determining how closely the second order kinetic equation fits the actual data points.

Kinetic analysis have yielded the 2^nd order rates of 1.17 ± 0.05 × 10⁶ M⁻¹s⁻¹ for repair of O⁶-POB-G paired with C and 1.74 ± 0.1 × 10⁶ M⁻¹S⁻¹ for O⁶-POB-G mispaired with T (Figure 7).

Discussion

The presence of functional AGT protein is crucial for cellular protection against alkylating agents such as nitrosamines present in tobacco smoke.^3,4 If not repaired, O⁶-alkylguanines can pair with a T or A instead of C during DNA replication, leading to G → A transitions and G → T transversion mutations.⁵ Human cancers with reduced expression levels of the AGT protein have an increased frequency of G → A mutations in the K-ras protooncogene and the p53 tumor suppressor gene as compared to tumors with normal expression of AGT,^29-32 indicative of a major protective function of AGT protein against these genetic changes. Epigenetic silencing of the MGMT gene coding for AGT has been correlated with susceptibility for tumor development following exposure to alkylating agents.³³

Previous studies have revealed that the efficiency of AGT-mediated repair of O⁶-alkylguanines is affected by adduct structure, local sequence context, and the identity of the base in the opposite strand.³⁴ The relative rates of AGT-mediated repair of O⁶-alkyl-dG lesions have been reported as benzyl > methyl > ethyl >> 2-hydroxyethyl >4-(3-pyridyl)-4-oxobutyl, probably due to a slower rate of transfer of bulky alkyl groups to the active site cysteine when they are bound in an inactive conformation.^14,34

AGT repair involves several distinct kinetic steps that have different rates.²⁸ Following binding to the minor groove of DNA, AGT protein flips the adducted nucleotide out of the DNA base stack to enter the protein active site pocket, while Arg128 takes the place of the adducted nucleotide in the DNA duplex.^6,9 The alkyl group is then transferred from the O⁶-position of guanine to the active site cysteine of AGT via an in-line displacement reaction, and finally, the alkylated AGT dissociates from the repaired DNA. The rate of AGT binding to DNA appears to be diffusion-limited (5 × 10⁹ M⁻¹ s⁻¹) and is unaffected by the identity of the alkyl group on the O⁶-alkyl-dG lesion or the nucleotide sequence surrounding the lesion,¹⁴ while the rate of protein dissociation is slower.²⁸ The rate of AGT-mediated nucleotide flipping is between 80 and 200 s^{−1 25} and is equally efficient for methyl and benzyl adducts.²⁸ The alkyl transfer step appears to be rate limiting for O⁶-Me-dG and O⁶-POB-dG adducts, although the rates of alkyl transfer have not been measured directly, but calculated from the rates for all of the other steps in the reaction.²⁸

In our earlier work, we have developed a quantitative HPLC ESI⁺-MS/MS methodology for analyzing the kinetics of AGT-mediated repair of O⁶-methyl-dG adducts in vitro.^19,25 Our analyses have employed second-order kinetics, as explained by a simple bimolecular reaction between the AGT protein and its DNA substrate (Scheme 1). Under these conditions, AGT repair of O⁶-methyl-dG exhibited a moderate sequence dependence, with ~2-fold differences between the rate of repair of O⁶-Me-dG located at the first and the second position of human K-ras codon 12.¹⁹ The dealkylation rate was weakly influenced by the methylation status of neighboring cytosine bases due to its effects on the rate of alkyl transfer.²⁵ While the overall effects of sequence context on O⁶-Me-dG repair by AGT were modest, due to the stochiometric nature of AGT reaction, even small differences in substrate preference may influence the probability of lesion repair at specific sites within the genome.⁸

In comparison to O⁶-Me-dG, AGT-mediated repair of bulky alkylguanine adducts such as O⁶-[4-oxo-4-(3-pyridyl)but-1-yl]deoxyguanosine (O⁶-POB-dG) appears to be more strongly affected by the local DNA sequence context.^13,14 As discussed above, O⁶-POB-dG is an important DNA lesion induced by tobacco-specific nitrosamines NNK and NNN and playing an important role in the carcinogenic mechanism of NNK.^4,5 In general, AGT repair of O⁶-POB-dG is less efficient than that of O⁶-Me-dG, probably due to the larger adduct size and the possibility of adduct binding to AGT active site in a non-reactive conformation.¹⁴ Coulter et al. and Mijal et al. previously reported that the first order rate for AGT repair of O⁶-POB-dG placed at the first guanine of H-ras codon 12 was greater as compared to adducts present at the second G.^13,14 However, to our knowledge, there has been no systematic study of the effects of DNA sequence on AGT-mediated repair of O⁶-POB-dG lesions.

In the present work, a non-radioactive, sensitive, and accurate isotope dilution HPLC-ESI⁺-MS/MS methodology was developed to analyze the kinetics of O⁶-POB-dG adduct repair by AGT. In our approach, DNA duplexes containing site-specific DNA adducts are incubated with recombinant human AGT protein for fixed periods of time, followed by acid hydrolysis and HPLC-ESI⁺-MS/MS analysis of O⁶-POB-G adducts remaining in DNA over time. Since no separation of alkylated and dealkylated strands is required, this methodology is highly reliable and is applicable to AGT repair studies with DNA of any sequence and length.

The applicability of the new method was demonstrated by analyzing the kinetics of AGT-mediated repair of O⁶-POB-dG lesions placed within the context of p53 codon 158 in the presence of increasing AGT amounts (Figure 5). As expected, the rate of O⁶-POB-dG repair increased proportionally as the AGT amounts were raised. Saturation of repair was observed when using low AGT:DNA ratios due to the depletion of active protein. As was the case with O⁶-Me-dG,¹⁹ we did not observe quantitative repair of O⁶-POB-dG lesions even when using 1.6 molar excess protein under these in vitro conditions, which could be a result of cooperative AGT-DNA binding¹⁰ and/or the competition of active protein with alkylated protein for binding to the DNA substrate.⁸

Second order rates of AGT-mediated repair of O⁶-POB-dG placed at the first position of rat H-ras codon 12 were determined with cytosine or thymine opposite the adduct. We found that the rate of AGT repair was 1.5 fold faster for the O⁶-POB-dG:T base pair than for the O⁶-POB-dG:C pair (1.74 ± 0.1 × 10⁶ M⁻¹s⁻¹ versus 1.17 ± 0.05 × 10⁶ M⁻¹s⁻¹) (Figure 7). This observation is consistent with an earlier report by Mijal et al.,¹³ who demonstrated that placing a thymine opposite a O⁶-POB-dG adduct caused an increase in AGT mediated repair as compared to O⁶-POB-dG:C base pair. By comparison, the rates of AGT-mediated repair of O⁶-Me-dG:C placed in a similar sequence context (K-ras codon 12) are 5-10 times greater (7.4 -14 × 10⁶ M⁻¹s⁻¹).^19,34 Our ongoing investigations (not shown) suggest that this difference in repair rates is due to a combination of slower nucleotide flipping and alkyl transfer steps in the case of bulky O⁶-POB-dG adducts.

In conclusion, an accurate and general isotope dilution capillary HPLC ESI⁺-MS/MS method has been developed to investigate the rates of AGT mediated repair of O⁶-POB-dG adducts as a function of local DNA sequence. This methodology will be useful for future investigations of the kinetics of AGT-mediated repair of pyridyloxobutylated DNA adducts.