Methyl-Cytosine-Driven Structural Changes Enhance Adduction Kinetics of an Exon 7 fragment of the p53 Gene

Methylation of cytosine (C) at C-phosphate-guanine (CpG) sites enhances reactivity of DNA towards electrophiles. Mutations at CpG sites on the p53 tumor suppressor gene that can result from these adductions are in turn correlated with specific cancers. Here we describe the first restriction-enzyme-assisted LC-MS/MS sequencing study of the influence of methyl cytosines (MeC) on kinetics of p53 gene adduction by model metabolite benzo[a]pyrene-7,8-dihydrodiol-9,10-epoxide (BPDE), using methodology applicable to correlate gene damage sites for drug and pollutant metabolites with mutation sites. This method allows direct kinetic measurements by LC-MS/MS sequencing for oligonucleotides longer than 20 base pairs (bp). We used MeC and non-MeC (C) versions of a 32 bp exon 7 fragment of the p53 gene. Methylation of 19 cytosines increased the rate constant 3-fold for adduction on G at the major reactive CpG in codon 248 vs. the non-MeC fragment. Rate constants for non-CpG codons 244 and 243 were not influenced significantly by MeC. Conformational and hydrophobicity changes in the MeC-p53 exon 7 fragment revealed by CD spectra and molecular modeling increase the BPDE binding constant to G in codon 248 consistent with a pathway in which preceding reactant binding greatly facilitates the rate of covalent SN2 coupling.

Scientific RepoRts | 7:40890 | DOI: 10.1038/srep40890 oligonucleotides and proposed enhanced subsequent coupling with MeCs related to stronger binding in a preceding step 18,19 . This model was supported by studies of alkyl-C exon 5 analogs 13 .
We recently reported a restriction-enzyme-assisted LC-MS/MS DNA sequencing method that extends reactive codon measurements to ds-DNA lengths longer than 20-bp 20 . For a 32 bp exon 7 p53 duplex fragment, we found adduct yields for reaction with BPDE in the order of codons 248 > 243 > 244. In the present paper, we adapt this method for the first time to elucidate relationships between codon-specific reaction kinetics with metabolites, and evaluate the influence of subtle but important MeC-related p53 gene structure changes on reaction kinetics. This approach enabling studies of longer nucleotides than have been previously possible is amenable to uncover the influence of important structural changes relevant to the reactivity of the entire gene.
We compared a MeC form (19 MeCs) with a non-MeC form ( Fig. 1b and c) of the 32-bp exon 7 p53 fragment. G in codon 248 CpG had the largest rate constant, which was 3-fold larger for the MeC version compared to all-C. Rate constants for reactive G's in non-CpG codons were 5-8 fold smaller than codon 248 MeCpG. Conformational and hydrophobicity changes in the MeC-p53 fragment revealed by circular dichroism (CD) and molecular modeling combine to increase the binding constant of BPDE at the codon 248 site to greatly facilitate the rate of the S N 2 coupling reaction in line with the preceding non-covalent binding pathway.

Results and Discussion
BPDE was reacted with an exon 7 fragment in solution, the BPDE-adducted oligonucleotide is cut by restriction enzyme NlAIII, then denatured by heat to obtain 4 single strand (ss) fragments ( Fig. 1b and c) that were analyzed by LC-MS/MS sequencing. MS/MS spectra of undamaged fragments were compared to that of singly adducted fragments to identify BPDE-adducted strands (Table S1, Supplementary Information (SI) file). The extracted  ion chromatogram (XIC) for singly adducted methylated ss-Fragment 2, m/z 1038.5 z = − 6 ( Fig. 2a) shows a single peak, indicating only one singly-adducted fragment 2 for the MeC exon 7 fragment. The MS/MS spectrum (Fig. 2b) shows a n -b n values similar to the undamaged MeC exon 7 up to a 5 -b 5 and an increase in m/z from a 6 -b 6 and above indicating adduction on the 5 th base. (AA Me C Me CG*GAGG Me C Me C Me CAT Me C Me CT Me CA, * = adduction site). This is confirmed by w ions similar to unreacted MeC fragment up to w 14 with an increase in w 15 (Table S2, SI file). Similar MS/MS spectra were analyzed for Fragment 1. (see SI. MS/MS spectra indicated two positional isomers for singly adducted fragment 1: ( Me CATGG*G Me CGGCATG) and ( Me CATG*GG Me CGGCATG).
MS/MS spectra obtained from the non-MeC version were similar to those reported in our previous study 20 . Codons selectively reacted with BPDE in MeC and C versions of the exon 7 fragment were 248, 244 and 243. As a control, BPDE was reacted with 32 bp exon 7 fragment of p53 having only one MeC adjacent to codon 245 while all others were un-methylated. No change in reaction selectivity was found.
Kinetics. Multiple reaction monitoring (MRM) was then used to quantify BPDE adduction at each reactive codon vs. reaction time. Transitions were selected specific to the ss-fragment monitored (Table S4, Figure S1b and c) which is also the major transition for the unadducted precursor ion, m/z 1009.9.
The relative amount of BPDE adduction was measured as ratio of peak area for XIC of adducted fragment to total peak area of the corresponding adducted + unadducted fragment. Relative amounts of BPDE adduction were plotted vs. time for G's in codons 248, 244 and 243 (Fig. 3a).
Expressions in eq (1) define k 1 as the pseudo-first order rate constant, and k 2 as the second order rate constant, where C o is initial amount of unreacted exon 7 fragment and C the amount unreacted at time t. Linear plots that fit eq 1 were obtained (Fig. 3b), as shown for codon 248 for MeC and all C exon 7 fragments. We obtained k 1 from the slopes, and k 2 from k 1 (equation 1). Plots for the other codons are in Figure S2, SI file.
Kinetic results show that k 2 for BPDE adduction on codon 248 CpG is nearly 3-fold larger for the MeC fragment vs. C-only (Fig. 3c, Table 1). Rate constants for non-CpG adduct at codons 244 and 243, are ~20% smaller for MeC vs C-only, but differences are not statistically significant. The k 2 ratio of the MeC oligonucleotide for  Structural Analysis. Additional studies were aimed at molecular interpretation of the kinetics. Circular dichroism (CD) spectra of full MeC and C-only versions of the exon 7 fragments (Fig. 4), suggest different conformations ( Figures S3, S4, SI for full analysis). The MeC exon 7 has an intense negative CD peak at 210 nm and an intense positive peak near 270 nm similar to a pure A-DNA structure 21 , but also a minimum near 245 nm characteristic of B-DNA. For the non-MeC exon 7, the first minimum is shifted to longer wavelength and is weaker, and a maximum at 265 nm is broad, with a shoulder at ~285 nm more characteristic of B DNA 21 (Fig. 4). We interpret both CD spectra in terms of mixed A-B DNA structures, with MeC's driving structure toward the A.

Molecular Modeling.
A and B forms of Me-C and C versions of exon 7 were constructed and molecular modeling was done using Autodock software. Structures were solvated with water and docked with the most reactive isomer (+ ) anti-BPDE. BPDE conformations at optimal docking sites were in the minor groove close to codon 248 for conformations with the most negative binding free energy. Conformations with distances between reactive exocyclic amine of G in codon 248 and the epoxide carbon of BPDE less 4.5 Å were considered, due to probability of subsequently forming covalent bonds. Optimal binding of BPDE to guanine in codon 248 (Fig. 5) gave binding free energies (∆ G b ) for B and A conformations ( Table 2) which were used to calculate binding constants (K b ) from K b = − ∆G b /RT, where R is the ideal gas constant and T is in Kelvin. Larger K b 's were found for MeC versions in both A and B form of DNA compared to all-C counterparts. For conformations approximating experimental ones, MeC A-form had 5-fold higher K b than all-C A-form. Smaller interatomic distances of reactive atoms were found in A-form of DNA (Fig. 5, Table 2) indicating better accessibility for BPDE, with the smallest distance for the MeC A-form. Docking studies were also done with (-) anti BPDE (SI file), which gave qualitatively similar results but less dramatic K b differences. As a control, we modeled BPDE binding to ds-poly(dG-dC).(dG-dC) oligonucleotides with all MeCs and all C, and found 5-fold larger K b for MeC version similar to experimental measurements 18 . Thus, modeling of the MeC exon 7 fragment as an A DNA structure and the all-C version closer to B DNA agreed well with a pathway featuring preceding non-covalent binding of BPDE in the minor groove near codon 248 that "sets up" subsequent fast S N 2 covalent coupling. It's very likely that stronger binding to the A DNA-like structure of the MeC exon 7 is also influenced considerably by hydrophobic interactions that increase for A DNA-like MeC oligonucleotides. An indirect indication of this effect was found in preliminary modeling studies without water, in which similar trends were found in K b for systems in Fig. 5, but K b differences between A MeC and B all-C forms were much smaller. We thus attribute a part of the increases in K b ( Table 2) for both the A and B forms of the MeC p53 fragments to the hydrophobic influence on water structure that tends to increase affinity of BPDE for the codon 248 minor groove.
Our findings of B-like to A-like structural changes for conversion from the C → MeC exon 7 duplexes are consistent with earlier literature 22,23 . Crystalized oligonucleotides are predominantly B DNA, but can transition to dehydrated A-forms upon methylation when rich in CG regions, and intermediate structures between A and B have been crystallized 24,25 . The A form has a wider minor groove that provides better accessibility for BPDE 21 . This enables a shorter distance for the reactive exocyclic amine of G to the epoxide carbon of BPDE in A-form than in B form ( Table 2). Earlier computations showed that epigenetic modifications alter the structure of the DNA making sites of adduction more accessible [26][27][28][29] .

Conclusions
Methods utilized above provide a straightforward approach to directly study kinetics of gene damage reactions. Results suggest that methylcytosines, which predominate in tumor suppressor genes 4 , and influence the kinetics of S N 2 reactions with BPDE mainly at CpG sites of tumor suppressor genes. In the p53 exon 7 fragment studied, codons 248, 244 and 243 were the reactive sites for MeC and all-C versions. Codon 248, the featuring CpG, gave the fastest reaction with the MeC fragment reacting 3-fold faster than the non-MeC version (Table 1). CD spectra and computation modeling uncovered a change in conformation from a mixed A-B to a more A-like duplex structure that drives free-energy for noncovalent binding of BPDE in the codon 248 region more negative for the MeC version, due to better access to the minor groove site and increased hydrophobicity. The resulting larger K b most likely lowers activation free energy to contribute significantly to the faster kinetics of S N 2 coupling of BPDE to MeCpG in codon 248. The structural change does not significantly influence non-CpG codons 244 and 243 that have similar kinetics for MeC and C versions. The hydrolysis-free methodology used here to measure direct kinetics of damage by metabolites to oligonucleotides longer than 20 bp is applicable to correlate gene damage sites for drug and pollutant metabolites with mutation sites. We speculate that these longer nucleotides are more amenable than shorter fragments to uncover the influence of important structural changes relevant to the reactivity of the entire gene.  Methodology described is directly adaptable to other chemicals and other tumor suppressor gene fragments to investigate kinetics of their DNA damage reactions. Molecular dynamics modeling can be used as an auxiliary tool to gain a more complete assessment of the chemistry of the associated reaction events. This quantitative methodology can be adapted to multiple chemicals and multiple exons across multiple tumor suppressor genes to expanding knowledge of genotoxicity chemistry pathways in relation to organ specificity of carcinogenesis.

Materials and Methods
Chemicals and Reagents. Benzo[a]pyrene-r-7,t-8-dihydrodiol-t-9,10-epoxide (± ) (anti) (anti-BPDE) was from National Cancer Institute Chemical Carcinogen Reference Standard Repository. Triethylammonium bicarbonate (1.0 M, pH 8.6), HPLC-grade methanol, acetonitrile and water were obtained from Sigma Aldrich. Custom made 32 base oligonucleotide fragments exon 7 p53 that were methylated on all C's (MeC) or had no MeCs were from Sigma Aldrich. All oligonucleotides were HPLC purified and the mass of non-MeC forward fragment is 9820, reverse fragment is 9833 and for the MeC version, forward strand has a mass of 9961 indicating the presence of 10 MeC and for reverse strand 9960 indicating the presence of 9 MeC. Restriction enzyme NlaIII was from New England Biolabs.
Reaction of Exon 7 Fragments with BPDE. 100 μ g (~5 nmol) of ds-32 base pair exon 7 fragment is reacted with 50 nmol of BPDE in a total reaction volume of 150 μ L, 10 mM Tris buffer pH 7.4 and 50 mM sodium chloride in the dark controlled at 25 ± 0.5 °C. This reaction was performed at various time intervals 2, 4, 6, 8, 12 and 24 hours for both unmethylated C and fully MeC versions of the ds-32 base pair exon 7 fragment. Reactions were stopped by adding cold acetonitrile and excess BPDE was removed using 3000 da molecular weight cut off filters (from EMD Millipore, UFC500396), which allows BPDE to pass through and DNA was collected from the filter as described previously 20 . These ds-32 base pair p53 gene fragments were subjected to restriction enzyme treatment and purification steps before subjected to LC-MS/MS analysis. All LC-MS/MS samples were run in triplicates. Restriction enzyme treatment on ds-32 base pair DNA. Approximately 100 μ g of ds-32 base pair was recovered from the reaction mixture and treated with 10 μ L (100 units) of NlaIII enzyme, 20 μ L of 10X NE buffer (from New England Biolabs) and the volume was made up to 200 μ L with water. The reaction mixture was incubated at 37 °C for 8 hours. DNA fragments were extracted from restriction enzyme reaction mixture using a previously described protocol using phenol/chloroform/isoamylalcohol, 25/24/1 and chloroform/isoamylalcohol, 24/1 to remove proteins. Briefly mixture of DNA and RE enzymes was vortexed with equal volume of phenol/ chloroform/isoamyl alcohol for 15 min followed by centrifugation for 10 min, organic phase discarded and aqueous phase was collected (for 3 times), and then repeated the same process with chloroform/isoamylalcohol (2 times). Finally the obtained DNA fragments from aqueous phase were subjected to desalting using Water's Oasis HLB cartridges (WAT094226) by solid phase extraction. Briefly cartridges were washed with methanol and water for equilibration followed by sample addition and washing the salts with 5% methanol and elution with 100% methanol. Obtained samples were evaporated in a rotovap, re-dissolved in water and heated and cooled to obtain ss fragments. Stored at − 20 °C until use.

Molecular modeling.
A and B form's of 32-base pair p53 DNA was modeled using make-na software 30 and modified with cytosines methylated using Maestro software and minimized 31 . Solvated models of these modified oligonucleotides were created using CHIMERA software 32,33 , Amber solvation model was used for solvation with a box size of 1 Å to accommodate water molecules. Autodock 4.2.6 was used for docking studies. Prepared biomolecule (Solvated MeC and C 32 base pair exon 7 fragment) were imported into the software. Lamarckian genetic algorithm (LGA) was used in Autodock 4.2.6 to find binding energy between the gene fragments and BPDE. Grid or volume for docking studies were kept constant for all the confirmations and set to be at maximum. Binding energies, binding constants and the distance between the exocyclic amine of the reactive guanine and epoxide carbon of BPDE were calculated 28 . Detailed steps for molecular modeling are given in SI.