Entropy involved in fidelity of DNA replication

Information has an entropic character which can be analyzed within the Statistical Theory in molecular systems. R. Landauer and C.H. Bennett showed that a logical copy can be carried out in the limit of no dissipation if the computation is performed sufficiently slowly. Structural and recent single-molecule assays have provided dynamic details of polymerase machinery with insight into information processing. We introduce a rigorous characterization of Shannon Information in biomolecular systems and apply it to DNA replication in the limit of no dissipation. Specifically, we devise an equilibrium pathway in DNA replication to determine the entropy generated in copying the information from a DNA template in the absence of friction. Both the initial state, the free nucleotides randomly distributed in certain concentrations, and the final state, a polymerized strand, are mesoscopic equilibrium states for the nucleotide distribution. We use empirical stacking free energies to calculate the probabilities of incorporation of the nucleotides. The copied strand is, to first order of approximation, a state of independent and non-indentically distributed random variables for which the nucleotide that is incorporated by the polymerase at each step is dictated by the template strand, and to second order of approximation, a state of non-uniformly distributed random variables with nearest-neighbor interactions for which the recognition of secondary structure by the polymerase in the resultant double-stranded polymer determines the entropy of the replicated strand. Two incorporation mechanisms arise naturally and their biological meanings are explained. It is known that replication occurs far from equilibrium and therefore the Shannon entropy here derived represents an upper bound for replication to take place. Likewise, this entropy sets a universal lower bound for the copying fidelity in replication.

💡 Research Summary

The paper presents a rigorous thermodynamic‑information‑theoretic analysis of DNA replication, focusing on the Shannon entropy associated with copying a template strand under the idealized condition of zero dissipation. Building on Landauer’s and Bennett’s principle that logical operations can be performed without energy loss if carried out quasistatically, the authors construct a hypothetical equilibrium pathway for polymerase activity. In this pathway the initial state consists of free deoxynucleoside triphosphates (dNTPs) randomly distributed in solution at known concentrations, while the final state is a newly polymerized strand that exactly mirrors the template. Both states are treated as mesoscopic equilibrium ensembles, allowing the use of Boltzmann statistics to assign probabilities to each possible nucleotide incorporation event.

The core of the analysis relies on experimentally measured nearest‑neighbor stacking free energies (ΔG) for all possible base‑pair steps. For a given template base, the probability that the polymerase incorporates a particular complementary dNTP is taken as proportional to exp(−ΔG/kT), where k is Boltzmann’s constant and T the absolute temperature. This yields a position‑specific probability distribution that can be interpreted as a set of random variables describing the replicated strand.

Two levels of approximation are explored. In the first‑order (independent‑site) model, each incorporation event is assumed to be statistically independent of its neighbors. The replicated strand is therefore modeled as a sequence of independent, non‑identically distributed random variables, and the total Shannon entropy H₁ is simply the sum of the site‑wise entropies. This model captures the basic template‑driven fidelity: the higher the free‑energy gap between the correct and incorrect base, the lower the local entropy and the higher the copying accuracy.

In the second‑order (nearest‑neighbor) model, the authors incorporate the well‑known stacking interactions between adjacent base pairs. Here the probability of inserting a nucleotide depends not only on the current template base but also on the identity of the previously incorporated nucleotide. Mathematically this is expressed as a first‑order Markov chain with transition matrix Pij ∝ exp(−ΔGij/kT). The total entropy H₂ is then H₁ minus the mutual information I(Xₙ;Xₙ₊₁) contributed by the correlations. This refinement acknowledges that the polymerase “senses” the emerging double‑helix geometry and that secondary structure influences the error landscape.

The authors emphasize that real cellular replication proceeds far from equilibrium, driven by the hydrolysis of nucleoside‑triphosphate bonds and by kinetic proofreading mechanisms. Consequently, the entropy calculated for the reversible equilibrium pathway constitutes an upper bound on the actual entropy production during replication. Because Shannon entropy sets a lower bound on the probability of error (ε ≥ 2⁻ᴴ), the derived H provides a universal thermodynamic limit on replication fidelity: no biological system can achieve an error rate lower than that dictated by the equilibrium free‑energy landscape.

Finally, the paper demonstrates how to use the model in practice. By inserting experimentally determined stacking ΔG values, the ambient temperature, and the concentrations of the four dNTPs, one can compute the site‑specific incorporation probabilities and thus predict the minimal achievable error rate for any given DNA sequence. This quantitative framework has immediate relevance for synthetic biology (design of high‑fidelity polymerases), DNA‑based data storage (optimizing error‑correction codes), and cancer biology (understanding how altered nucleotide pools or polymerase mutations shift the thermodynamic error floor).

In summary, the study bridges statistical physics, information theory, and molecular biology to define a thermodynamic ceiling for the entropy generated during DNA copying and a corresponding floor for replication accuracy. It shows that even in the hypothetical limit of zero dissipation, the intrinsic free‑energy differences among base‑pair steps impose a non‑zero informational cost, thereby setting universal constraints on the fidelity of life’s most fundamental copying process.