The future of biomolecular simulation in the pharmaceutical industry: what we can learn from aerodynamics modelling and weather prediction. Part 1. understanding the physical and computational complexity of in silico drug design.

The predictive power of simulation has become embedded in the infrastructure of modern economies. Computer-aided design is ubiquitous throughout industry. In aeronautical engineering, built infrastructure and materials manufacturing, simulations are routinely used to compute the performance of potential designs before construction. The ability to predict the behaviour of products is a driver of innovation by reducing the cost barrier to new designs, but also because radically novel ideas can be piloted with relatively little risk. Accurate weather forecasting is essential to guide domestic and military flight paths, and therefore the underpinning simulations are critical enough to have implications for national security. However, in the pharmaceutical and biotechnological industries, the application of computer simulations remains limited by the capabilities of the technology with respect to the complexity of molecular biology and human physiology. Over the last 30 years, molecular-modelling tools have gradually gained a degree of acceptance in the pharmaceutical industry. Drug discovery has begun to benefit from physics-based simulations. While such simulations have great potential for improved molecular design, much scepticism remains about their value. The motivations for such reservations in industry and areas where simulations show promise for efficiency gains in preclinical research are discussed. In this, the first of two complementary papers, the scientific and technical progress that needs to be made to improve the predictive power of biomolecular simulations, and how this might be achieved, is firstly discussed (Part 1). In Part 2, the status of computer simulations in pharma is contrasted with aerodynamics modelling and weather forecasting, and comments are made on the cultural changes needed for equivalent computational technologies to become integrated into life-science industries.

Single-molecule fluorescence resonance energy transfer assays reveal heterogeneous folding ensembles in a simple RNA stem-loop.

We have examined the folding ensembles present in solution for a series of RNA oligonucleotides that encompass the replicase translational operator stem-loop of the RNA bacteriophage MS2. Single-molecule (SM) fluorescence assays suggest that these RNAs exist in solution as ensembles of differentially base-paired/base-stacked states at equilibrium. There are two distinct ensembles for the wild-type sequence, implying the existence of a significant free energy barrier between "folded" and "unfolded" ensembles. Experiments with sequence variants are consistent with an unfolding mechanism in which interruptions to base-paired duplexes, in this example by the single-stranded loop and a single-base bulge in the base-paired stem, as well as the free ends, act as nucleation points for unfolding. The switch between folded and unfolded ensembles is consistent with a transition that occurs when all base-pairing and/or base-stacking interactions that would orientate the legs of the RNA stem are broken. Strikingly, a U-to-C replacement of a residue in the loop, which creates a high-affinity form of the operator for coat protein binding, results in dramatically different (un)folding behaviour, revealing distinct subpopulations that are either stabilised or destabilised with respect to the wild-type sequence. This result suggests additional reasons for selection against the C-variant stem-loop in vivo and provides an explanation for the increased affinity.

Modelling the biomechanical properties of DNA using computer simulation.

Duplex DNA must remain stable when not in use to protect the genetic material. However, the two strands must be separated whenever genes are copied or expressed to expose the coding strand for synthesis of complementary RNA or DNA bases. Therefore, the double stranded structure must be relatively easy to take apart when required. These conflicting biological requirements have important implications for the mechanical properties of duplex DNA. Considerable insight into the forces required to denature DNA has been provided by nanomanipulation experiments, which measure the mechanical properties of single molecules in the laboratory. This paper describes recent computer simulation methods that have been developed to mimic nanomanipulation experiments and which, quite literally, 'destruction test' duplex DNA in silico. The method is verified by comparison with single molecule stretching experiments that measure the force required to unbind the two DNA strands. The model is then extended to investigate the thermodynamics of DNA bending and twisting. This is of biological importance as the DNA must be very tightly packaged to fit within the nucleus, and is therefore usually found in a highly twisted or supercoiled state (in bacteria) or wrapped tightly around histone proteins into a densely compacted structure (in animals). In particular, these simulations highlight the importance of thermal fluctuations and entropy in determining the biomechanical properties of DNA. This has implications for the action of DNA processing molecular motors, and also for nanotechnology. Biological machines are able to manipulate single molecules reliably on an energy scale comparable to that of thermal noise. The hope is that understanding the statistical mechanisms that a cell uses to achieve this will be invaluable for the future design of 'nanoengines' engineered to perform new technological functions at the nanoscale.