4.4 Conclusions

The unzipping of DNA at controlled position exhibits cooperative openings of base-pairs (called CURs) as the distance between the two beads is increased. These CURs show characteristic statistical properties that depend on the sequence of the DNA and the parameters of the experiments (e.g., trap stiffness).

We have developed a Bayesian approach to extract the statistical properties of CURs directly from the experimental data. Such analysis can be extended to other systems (such as proteins or RNA) that show intermediate states in the unfolding process. The obtained histogram of intermediate states is a sign of the molecule and contains valuable information about the stability and the coexistence of these metastable states.

The properties of the CURs can be described by a simplified toy model that captures the essential features of the unzipping. The advantage of the toy model is that the calculations are easy to implement and the computational time required for the calculations is dramatically reduced. This is useful to intensively explore the predictions of the unzipping properties for different situations: trap stiffness, NNBP variances, sequences, etc. The results predicted by the toy model compare well with the experimental data obtained from unzipping.

The detection of intermediate states using the Bayesian analysis is blurred by the thermal noise of the experimental data. In our experimental conditions, the CURs smaller than $ \sim$10 bp are hardly detected. Below this boundary, the fluctuations in force due to the thermal noise cannot be distinguished from the fluctuations due to the coexistence of states. A priori, this limit could be reduced down to 1 bp by collecting more accurate experimental data and finding the characteristic signature of each type of noise (correlation, spectrum, bandwidth).

The CUR size of the experimental unzipping is affected by the release of ssDNA. As the molecule is being unzipped, the amount of ssDNA between the unzipping fork and the optical trap increases. These reduces the compliance of the tether and the capability to distinguish between intermediate states is less clear. The unzipping of a DNA molecule one base-pair at a time requires a minimum stiffness of 100 pN/nm ($ =0.1$ N/m). This could be achieved by applying local force on the unzipping fork, avoiding the accumulation of ssDNA that transmits the force to the optical trap.

Interestingly enough, the minimum stiffness required to unzip one base-pair at a time coincides with the expected stiffness of one single base of the ssDNA. On the other hand, the stiffness of the proteins that directly pull on DNA (e.g., helicases) or read the sequence (e.g., polymerases) can be assumed to be very large (proteins are very rigid objects) compared to the stiffness of a single base-pair. Therefore, the stiffness of the proteins are not a limiting factor to unzip DNA molecules one base-pair at a time. It is remarkable that ssDNA has the minimum required elastic properties so that the genetic information encoded in DNA can be accessed by the replication and transcription machinery.

To sum up, the unzipping of DNA one base-pair at a time could be used to infer the sequence of an unknown DNA fragment. This could be experimentally achieved by increasing the stiffness of the probe, applying local force to the unzipping fork and improving the Bayesian analysis to distinguish the thermal force fluctuations from the force fluctuations due to the coexistence of states.

JM Huguet 2014-02-12