This is a brief overview of my current work on a protocol for studying molecular mechanisms in biomolecules. It is based on Natural Move Monte Carlo (NMMC), a technique pioneered by one of my supervisors, Peter Minary.
NMMC allows the user to decompose a protein or nucleic acid structure into segments of atoms/residues. Each segment is moved collectively and treated as a single body. This gives rise to a sampling strategy that considers the structure as a fluid of segments and probes the different arrangements in a Monte Carlo fashion.
Traditionally the initial decomposition would be the only one that is sampled. However, this decomposition might not always be optimal. Critical degrees of freedom might have been missed out or chosen sub-optimally. Additionally, if we want to test the causality of a structural mechanism it can be informative to perform NMMC simulations for a variety of decompositions. Here I show an example of how customised Natural Moves may be applied on a DNA system.
Investigating the effect of epigenetic marks on the structure of a DNA toy model
Figure 1. Three levels at which epigenetic activity takes place. Figure adopted from Charles C. Matouk, and Philip A. Marsden Circulation Research. 2008;102:873-887
Epigenetic marks on DNA nucleotides are involved in regulating gene expression (Point 1 in figure 1). We have a limited understanding of the underlying molecular mechanism of this process. There are two mechanisms that are thought to be involved: 1) Direct recognition of the epigenetic mark by DNA binding proteins and 2) indirect recognition of changes in the local DNA structure caused by the epigenetic mark. Using customised Natural Moves we are currently trying to to gain insight into these mechanisms.
One type of epigenetic mark is the 5-hydroxymethylation (5hm) on cytosines. Lercher et al. have recently solved a crystal structure of the Drew-Dickerson Dodecamer with two of these epigenetic marks attached. Given the right sequence context (e.g. CpG) they have found that this epigenetic mark can form a hydrogen bond between two neighbouring bases on the same strand. This raises the question: Can an intra-strand CpG hydrogen bond alter the DNA helical parameters? We used this as a toy model to test our technology.
Figure 2b
Figure 2a
Figure 2a shows the Drew-Dickerson Dodecamer schematically. Figure 2b shows the three sets of degrees of freedom that we used as test cases.
Top (11): Default sampling – Serves as a reference simulation.
Middle (01): Fixed 5hm torsions – By forcing the hydroxyl group of 5hmC towards the guanine we significantly increase the chance of the hydrogen bond forming.
Bottom (00): Collective movement of the neighbouring bases 5hmC and G – By grouping the two bases into a segment we aim to emulate the dampening effect that a hydrogen bond may have on their movements relative to each other.
By modifying these degrees of freedom (1=active/ 0=inactive) we attempted to amplify any effects that the CpG hydrogen bond may have on the DNA structure.
Below you can see animations of the three test cases 11, 01, 00 during simulation, zoomed in on the 5hm modification and the neighbouring base pair:
It appeared that the default degrees of freedom were not sufficient to detect a change in the DNA structure when comparing simulations of unmodified and modified structures. The other two test cases with customised Natural Moves, however, showed alterations in some of the helical parameters.
Lercher et al. saw no differences in their modified and unmodified crystal structures. It seems that single 5hm epigenetic marks are not sufficient to significantly alter DNA structure. Rather, clusters of these modifications with accumulated structural effects may be required to cause significant changes in DNA helical parameters. CpG islands may be a promising candidate.
