Skip to main content

Making microarrays rational - computational prediction of DNA duplex stability

Final Activity Report Summary - DUPLEX (Making microarrays rational - computational prediction of DNA duplex stability)

One of the most important examples of molecular self-assembly is the carrier of the genetic code, deoxyribonucleic acid (DNA). Non-covalent interactions are pivotal in self-assembly. These interactions are responsible for holding together helical DNA strings, each of which consists of a polymer of natural nucleic acids, namely guanine (G), thymine (T), cytosine(C) and adenosine (A). Hydrogen bonding is a prime concept, partially responsible for stabilising the interaction between DNA strings. Watson-Crick hydrogen-bonded base pairs are said to be held together by hydrogen bonds. On the other hand, the so-called pi-pi stacking, which is governed by dispersion, stabilises the interaction between nucleic acids within the same string.

Fully understanding the forces that keep DNA together is a complex task that continues to receive attention from both experiment and theory and computation. On the computational side, ab initio computations help in disentangling these forces. However, due to the great computational cost of these calculations, one must work with fragments of DNA. A meaningful fragment is just 'one rung of the DNA ladder', that is, G interacting with C (G...C) or A interacting with T (A...T). Chemical substitution enables to alter the interaction energy of a base pair in a systematic way, therefore providing insight into what governs the stability of hydrogen-bonded base pairs.

In one of our studies the nucleic acid C was substituted in the 5-position (denoted C5X) by 38 different functional groups. We found a remarkably linear correlation between the interaction energies of the various substituted base pairs and certain molecular properties of cytosine itself. These properties were various quantum chemical functions that were evaluated at the so-called bond critical points. The latter were special points, defined by the theory of quantum chemical topology, which could be located with minimal computational cost. A simple equation was proposed and properly scrutinised statistically, predicting the interaction energy of C5X...G from very few properties of cytosine. The advantage was that when new substituents were introduced, only information of the monomer (C5X) had to be computed. This saved enormous amounts of computation time.

The fact that such a linear relationship existed was remarkable, more so because we confirmed it for cytosine substituted in the 6 position (C6X) and for guanine in the 8 position (G8X). The bond critical point descriptors were superior to more familiar measures such as the Hammett constant and Mulliken charges. Systematic substitution of a stacked system of C and G led to a correlation coefficient r2 of 0.8. This once again showed the existence of hidden relationships. In this case they demonstrated quantitatively that an electron donating group on G and an electron withdrawing group on C increased the interaction energy.

In summary, we contributed to the difficult problem of what held a DNA string together. Our approach, although rooted in computation, adopted the philosophy of physical organic chemistry, a combination that appeared to be new and fruitful.