DNA is continuously exposed to a vast number of damaging events triggered by endogenous and exogenous agents. Numerous experimental studies have provided very useful information regarding structural properties of some of the DNA lesions and their repair. However, they have not supplied any mechanistic or energetic information pertaining to their formation. Computational Biochemistry has recently beneficiated from the advent of large scale  methods, which rely on the increase of available, highly-parallelized computational resources, but were first and foremost motivated by the utter need to describe as accurately as possible complex, strongly heterogeneous systems. Such multiscale and dynamical simulations have become routine for proteic systems, yet remain scarce on DNA. QM/MM methodologies are particularly useful in this regard, as they can be used to study reaction mechanisms and electronic properties of large systems.

In this thesis, we study oxidatively generated crosslinks within DNA.  Their formation poses a potent threat to genome integrity, because of their high mutagenic frequency. Many aspects about their formation and their structural impact on the regular B-DNA conformation are still unknown. In absence of NMR or X-ray structures, the B-helix distortion can be inspected purely on the basis of molecular simulations. We study the formation of both intra- and inter-strand cross links. The effect of DNA structure on their formation is discussed. 

The first class of crosslinks discussed in this study is the Purine-Pyrimidine intra-strand crosslinks. These crosslinks are generated through formation of a covalent linkage between a pyrimidine radical, formed due to H abstraction by hydroxyl radical, and a vicinal purinic base. We study energetics for the coupling of three such crosslinks, namely G[8--5Me]T, G[8-5]C and G[8-5Me]mC, by varying the pyrimidinic bases embedded in a specific DNA sequence using Car-Parrinello Molecular Dynamics (CPMD) within QM/MM framework. Our simulations qualitatively agree with the observed reactivity of the pyrimidinic bases towards guanine (T^● > mC^● > C^●) within B-DNA environment. Interestingly, this order of reactivity is reversed  for isolated nuclobases elucidating the crucial role of B-helix environment. Further, our data also suggests a severe distortion of B-helix for crosslink between guanine and cytosine.

The second class of crosslinks that we study pertains to formation of inter-strand crosslinks within DNA emanating from the product of C4ʹ oxidation of sugar moiety. The cascade of reaction subsequent to the abstraction of 4ʹ H atom from the sugar moiety leads to formation of a reactive conjugate keto-aldehyde accompanied with a single strand break. The conjugate keto-aldehyde can then react with nucleophillic DNA bases viz. Adenine, Cytosine or Guanine on the opposite strand leading to formation of inter-strand crosslinks. The single strand break imparts a high degree of conformational flexibility to DNA and thereby prohibiting the use of computationally expensive approaches like CPMD. Thus, we carry out our study in two parts. First, we study the reactivity of α,β-unsaturated keto-aldehydes with the three nucleophillic bases present in DNA using static QM methodologies. This allows us to elucidate the correct mechanism for the reaction including the stereochemistry of the products. The reactivity order predicted from our static calculation (C > A > G) is in qualitative agreement with experiments on isolated nucleobases. In addition, since other structurally similar conjugate keto-aldehydes are known mutagens, this exercise helps us to comment on their reactivity as well.  Second, we study the impact of the single strand break on the DNA structure, using classical MD, by incorporating the keto-aldehyde and the nucleobase opposite to it (C, A or G) within a poly-GC DNA sequence. Our results indicate that the keto-aldehyde and the opposing nucleobase can remain in proximity due to hydrogen bonding interaction. The amount of time keto-aldehyde and the opposing nucleobase remain in proximity is in the order C > G > A. Further, the overall structure of the DNA after strand break is impacted by the nature of nucleobase opposite to keto-aldehyde. Purinic bases show similar overall B-DNA structure with low bend and high twist angle values around the strand break, whereas, the pyrimidine base displays high bend and low twist structure.

KEYWORDS : DNA Damage, DNA cross-links, QM/MM, Molecular Dynamics, DFT, CPMD.