Mesoscopic model parametrization of hydrogen bonds and stacking interactions of RNA from melting temperatures

Weak-cooperative binding of a long single-stranded DNA chain on a surface

Binding gene-wide single-stranded nucleic acids to surface-immobilized complementary probes is an important but challenging process for biophysical studies and diagnostic applications. The challenge comes from the conformational dynamics of the long chain that affects its accessibility and weakens its hybridization to the probes. We investigated the binding of bacteriophage genome M13mp18 on several different 20-mer probes immobilized on the surface of a multi-spot, label-free biosensor, and observed that only a few of them display strong binding capability with dissociation constant as low as 10 pM. Comparing experimental data and computational analysis of the M13mp18 chain structural features, we found that the capturing performance of a specific probe is directly related to the multiplicity of binding sites on the genomic strand, and poorly connected with the predicted secondary and tertiary structure. We show that a model of weak cooperativity of transient bonds is compatible with the measured binding kinetics and accounts for the enhancement of probe capturing observed when more than 20 partial pairings with binding free energy lower than -10 kcal mol-1 are present. This mechanism provides a specific pattern of response of a genomic strand on a panel of properly selected oligomer probe sequences.

Atomistic simulations of RNA duplex thermal denaturation: Sequence- and forcefield-dependence

Double-stranded RNA is the end-product of template-based replication, and is also the functional state of some biological RNAs. Similarly to proteins and DNA, they can be denatured by temperature, with important physiological and technological implications. Here, we use an in silico strategy to probe the thermal denaturation of RNA duplexes. Following previous results that were obtained on a few different duplexes, and which nuanced the canonical 2-state picture of nucleic acid denaturation, we here specifically address three different aspects that greatly improve our description of the temperature-induced dsRNA separation. First, we investigate the effect of the spatial distribution of weak and strong base-pairs among the duplex sequence. We show that the deviations from the two-state dehybridization mechanism are more pronounced when a strong core is flanked with weak extremities, while duplexes with a weak core but strong extremities exhibit a two-state behavior, which can be explained by the key role played by base fraying. This was later verified by generating artificial hairpin or circular states containing one or two locked duplex extremities, which results in an important reinforcement of the entire HB structure of the duplex and higher melting temperatures. Finally, we demonstrate that our results are little sensitive to the employed combination of RNA and water forcefields. The trends in thermal stability among the different sequences as well as the observed unfolding mechanisms (and the deviations from a two-state scenario) remain the same regardless of the employed atomistic models. However, our study points to possible limitations of recent reparametrizations of the Amber RNA forcefield, which sometimes results in duplexes that readily denature under ambient conditions, in contradiction with available experimental results.

Twist-stretch relations in nucleic acids

Nucleic acids are highly deformable helical molecules constantly stretched, twisted and bent in their biological functioning. Single molecule experiments have shown that double stranded (ds)-RNA and standard ds-DNA have opposite twist-stretch patterns and stretching properties when overwound under a constant applied load. The key structural features of the A-form RNA and B-form DNA helices are here incorporated in a three-dimensional mesoscopic Hamiltonian model which accounts for the radial, bending and twisting fluctuations of the base pairs. Using path integral techniques which sum over the ensemble of the base pair fluctuations, I compute the average helical repeat of the molecules as a function of the load. The obtained twist-stretch relations and stretching properties, for short A- and B-helical fragments, are consistent with the opposite behaviors observed in kilo-base long molecules.

Cation valence dependence of hydrogen bond and stacking potentials in DNA mesoscopic models

Monovalent and divalent cations play a crucial role in living cells and for molecular techniques such as PCR. Here we evaluate DNA melting temperatures in magnesium (Mg²⁺) and magnesium‑potassium (Mg²⁺+ K⁺) buffers with a mesoscopic model that allows us to estimate hydrogen bonds and stacking interaction potentials. The Mg²⁺ and Mg²⁺+ K⁺ results are compared to previous calculations for sodium ions (Na⁺), in terms of equivalent sodium concentration and ionic strength. Morse potentials, related to hydrogen bonding, were found to be essentially constant and unaffected by cation conditions. However, for stacking interactions we find a clear dependence with ionic strength and cation valence. The highest ionic strength variations, for both hydrogen bonds and stacking interactions, was found at the sequence terminals. This suggests that end-to-end interactions in DNA will be strongly dependent on cation valence and ionic strength.

DNA-Based Biosensors for the Biochemical Analysis: A Review

In recent years, DNA-based biosensors have shown great potential as the candidate of the next generation biomedical detection device due to their robust chemical properties and customizable biosensing functions. Compared with the conventional biosensors, the DNA-based biosensors have advantages such as wider detection targets, more durable lifetime, and lower production cost. Additionally, the ingenious DNA structures can control the signal conduction near the biosensor surface, which could significantly improve the performance of biosensors. In order to show a big picture of the DNA biosensor's advantages, this article reviews the background knowledge and recent advances of DNA-based biosensors, including the functional DNA strands-based biosensors, DNA hybridization-based biosensors, and DNA templated biosensors. Then, the challenges and future directions of DNA-based biosensors are discussed and proposed.

Base pair fluctuations in helical models for nucleic acids

A statistical method is developed to estimate the maximum amplitude of the base pair fluctuations in a three dimensional mesoscopic model for nucleic acids. The base pair thermal vibrations around the helix diameter are viewed as a Brownian motion for a particle embedded in a stable helical structure. The probability to return to the initial position is computed, as a function of time, by integrating over the particle paths consistent with the physical properties of the model potential. The zero time condition for the first-passage probability defines the constraint to select the integral cutoff for various macroscopic helical conformations, obtained by tuning the twist, bending, and slide motion between adjacent base pairs along the molecule stack. Applying the method to a short homogeneous chain at room temperature, we obtain meaningful estimates for the maximum fluctuations in the twist conformation with ∼10.5 base pairs per helix turn, typical of double stranded DNA helices. Untwisting the double helix, the base pair fluctuations broaden and the integral cutoff increases. The cutoff is found to increase also in the presence of a sliding motion, which shortens the helix contour length, a situation peculiar of dsRNA molecules.

Complete Mesoscopic Parameterization of Single LNA Modifications in DNA Applied to Oncogene Probe Design

The use of mesoscopic models to describe the thermodynamic properties of locked nucleic acid (LNA)-modified nucleotides can provide useful insights into their properties, such as hydrogen-bonding and stacking interactions. In addition, the mesoscopic parameters can be used to optimize LNA insertion in probes, to achieve accurate melting temperature predictions, and to obtain duplex opening profiles at the base-pair level. Here, we applied this type of model to parameterize a large set of melting temperatures for LNA-modified sequences, from published sources, covering all possible nearest-neighbor configurations. We have found a very large increase in Morse potentials, which indicates very strong hydrogen bonding as the main cause of improved LNA thermodynamic stability. LNA-modified adenine-thymine (AT) was found to have similar hydrogen bonding to unmodified cytosine-guanine (CG) base pairs, while for LNA CG, we found exceptionally large hydrogen bonding. In contrast, stacking interactions, which were thought to be behind the stability of LNA, were similar to unmodified DNA in most cases. We applied the new LNA parameters to the design of BRAF, KRAS, and EGFR oncogene variants by testing all possible LNA modifications. Selected sequences were then synthesized and had their hybridization temperatures measured, achieving a prediction accuracy within 1 °C. We performed a detailed base-pair opening analysis to discuss specific aspects of these probe hybridizations that may be relevant for probe design.

Salt dependent mesoscopic model for RNA at multiple strand concentrations

Mesoscopic models can be used for the description of the thermodynamic properties of RNA duplexes. With the use of experimental melting temperatures, its parametrization can provide important insights into its hydrogen bonds and stacking interactions as has been done for high sodium concentrations. However, the RNA parametrization for lower salt concentrations is still missing due to the limited amount of published melting temperature data. While the Peyrard-Bishop (PB) parametrization was found to be largely independent of strand concentrations, it requires that all temperatures are provided at the same strand concentrations. Here we adapted the PB model to handle multiple strand concentrations and in this way we were able to make use of an experimental set of temperatures to model the hydrogen bond and stacking interactions at low and intermediate sodium concentrations. For the parametrizations we make a distinction between terminal and internal base pairs, and the resulting potentials were qualitatively similar as we obtained previously for DNA. The main difference from DNA parameters, was the Morse potentials at low sodium concentrations for terminal r(AU) which is stronger than d(AT), suggesting higher hydrogen bond strength.

RNA Nanoparticles as Rubber for Compelling Vessel Extravasation to Enhance Tumor Targeting and for Fast Renal Excretion to Reduce Toxicity

Rubber is a fascinating material in both industry and daily life. The development of elastomeric material in nanotechnology is imperative due to its economic and technological potential. By virtue of their distinctive physicochemical properties, nucleic acids have been extensively explored in material science. The Phi29 DNA packaging motor contains a 3WJ with three angles of 97°, 125°, and 138°. Here, the rubber-like property of RNA architectures was investigated using optical tweezers and in vivo imaging technologies. The 3WJ 97° interior angle was contracted or stretched to 60°, 90°, and 108° at will to build elegant RNA triangles, squares, pentagons, cubes, tetrahedrons, dendrimers, and prisms. RNA nanoarchitecture was stretchable and shrinkable by optical tweezer with multiple extension and relaxation repeats like a rubber. Comparing to gold and iron nanoparticles with the same size, RNA nanoparticles display stronger cancer-targeting outcomes, while less accumulation in healthy organs. Generally, the upper limit of renal excretion is 5.5 nm; however, the 5, 10, and 20 nm RNA nanoparticles passed the renal filtration and resumed their original structure identified in urine. These findings solve two previous mysteries: (1) Why RNA nanoparticles have an unusually high tumor targeting efficiency since their rubber or amoeba-like deformation property enables them to squeeze out of the leaky vasculature to improve the EPR effect; and (2) why RNA nanoparticles remain non-toxic since they can be rapidly cleared from the body via renal excretion into urine with little accumulation in the body. Considering its controllable shape and size plus its rubber-like property, RNA holds great promises for industrial and biomedical applications especially in cancer therapeutics delivery.

Optical and theoretical study of strand recognition by nucleic acid probes

Detection of nucleic acids is crucial to the study of their basic properties and consequently to applying this knowledge to the determination of pathologies such as cancer. In this work, our goal is to determine new trends for creating diagnostic tools for cancer driver mutations. Herein, we study a library of natural and modified oligonucleotide duplexes by a combination of optical and theoretical methods. We report a profound effect of additives on the duplexes, including nucleic acids as an active crowder. Unpredictably and inconsistent with DNA+LNA/RNA duplexes, locked nucleic acids contribute poorly to mismatch discrimination in the DNA+LNA/DNA duplexes. We develop a theoretical framework that explains poor mismatch discrimination in KRAS oncogene. We implement our findings in a bead-bait genotyping assay to detect mutated human cancer RNA. The performance of rationally designed probes in this assay is superior to the LNA-primer polymerase chain reaction, and it agrees with sequencing data.

Melting temperature measurement and mesoscopic evaluation of single, double and triple DNA mismatches

Unlike the canonical base pairs AT and GC, the molecular properties of mismatches such as hydrogen bonding and stacking interactions are strongly dependent on the identity of the neighbouring base pairs. As a result, due to the sheer number of possible combinations of mismatches and flanking base pairs, only a fraction of these have been studied in varying experiments or theoretical models. Here, we report on the melting temperature measurement and mesoscopic analysis of contiguous DNA mismatches in nearest-neighbours and next-nearest neighbour contexts. A total of 4032 different mismatch combinations, including single, double and triple mismatches were covered. These were compared with 64 sequences containing all combinations of canonical base pairs in the same location under the same conditions. For a substantial number of single mismatch configurations, 15%, the measured melting temperatures were higher than the least stable AT base pair. The mesoscopic calculation, using the Peyrard-Bishop model, was performed on the set of 4096 sequences, and resulted in estimates of on-site and nearest-neighbour interactions that can be correlated to hydrogen bonding and base stacking. Our results confirm many of the known properties of mismatches, including the peculiar sheared stacking of tandem GA mismatches. More intriguingly, it also reveals that a number of mismatches present strong hydrogen bonding when flanked on both sites by other mismatches. To highlight the applicability of our results, we discuss a number of practical situations such as enzyme binding affinities, thymine DNA glycosylase repair activity, and trinucleotide repeat expansions.

An asymmetric mesoscopic model for single bulges in RNA

Simple one-dimensional DNA or RNA mesoscopic models are of interest for their computational efficiency while retaining the key elements of the molecular interactions. However, they only deal with perfectly formed DNA or RNA double helices and consider the intra-strand interactions to be the same on both strands. This makes it difficult to describe highly asymmetric structures such as bulges and loops and, for instance, prevents the application of mesoscopic models to determine RNA secondary structures. Here we derived the conditions for the Peyrard-Bishop mesoscopic model to overcome these limitations and applied it to the calculation of single bulges, the smallest and simplest of these asymmetric structures. We found that these theoretical conditions can indeed be applied to any situation where stacking asymmetry needs to be considered. The full set of parameters for group I RNA bulges was determined from experimental melting temperatures using an optimization procedure, and we also calculated average opening profiles for several RNA sequences. We found that guanosine bulges show the strongest perturbation on their neighboring base pairs, considerably reducing the on-site interactions of their neighboring base pairs.

A nucleotide-level coarse-grained model of RNA

We present a new, nucleotide-level model for RNA, oxRNA, based on the coarse-graining methodology recently developed for the oxDNA model of DNA. The model is designed to reproduce structural, mechanical, and thermodynamic properties of RNA, and the coarse-graining level aims to retain the relevant physics for RNA hybridization and the structure of single- and double-stranded RNA. In order to explore its strengths and weaknesses, we test the model in a range of nanotechnological and biological settings. Applications explored include the folding thermodynamics of a pseudoknot, the formation of a kissing loop complex, the structure of a hexagonal RNA nanoring, and the unzipping of a hairpin motif. We argue that the model can be used for efficient simulations of the structure of systems with thousands of base pairs, and for the assembly of systems of up to hundreds of base pairs. The source code implementing the model is released for public use.

Mesoscopic model and free energy landscape for protein-DNA binding sites: analysis of cyanobacterial promoters

The identification of protein binding sites in promoter sequences is a key problem to understand and control regulation in biochemistry and biotechnological processes. We use a computational method to analyze promoters from a given genome. Our approach is based on a physical model at the mesoscopic level of protein-DNA interaction based on the influence of DNA local conformation on the dynamics of a general particle along the chain. Following the proposed model, the joined dynamics of the protein particle and the DNA portion of interest, only characterized by its base pair sequence, is simulated. The simulation output is analyzed by generating and analyzing the Free Energy Landscape of the system. In order to prove the capacity of prediction of our computational method we have analyzed nine promoters of Anabaena PCC 7120. We are able to identify the transcription starting site of each of the promoters as the most populated macrostate in the dynamics. The developed procedure allows also to characterize promoter macrostates in terms of thermo-statistical magnitudes (free energy and entropy), with valuable biological implications. Our results agree with independent previous experimental results. Thus, our methods appear as a powerful complementary tool for identifying protein binding sites in promoter sequences.

Binding of specific proteins to particular sites in the DNA sequence is a fundamental issue for gene regulation in molecular biology and genetic engineering. A deep understanding of cell physiology requires the analysis of a plethora of genes involving characterization of their promoter architectures that determine their regulation and gene transcription. In order to locate the promoter elements of a given gene, experimental determination of its transcription start site (TSS) is required. This is an expensive, time-consuming task that, depending on our requirements, could be simplified using computational analysis as a first approach. Nevertheless, most computational methods lack a physical basis on the protein-DNA interaction mechanism. We adopt here this strategy, by using a simple model for protein-DNA interaction to find TSS in a bunch of cyanobacteria promoters. We make use of physical tools to characterize these TSS and to relate them with biological properties as the relative strength of the promoter. Our study shows how a model based on a coarse-grained description of a biomolecule can give valuable insight on its biological function.

Twisting and bending stress in DNA minicircles

The interplay between bending of the molecule axis and appearance of disruptions in circular DNA molecules, with ∼100 base pairs, is addressed. Three minicircles with different radii and almost equal contents of AT and GC pairs are investigated. The DNA sequences are modeled by a mesoscopic Hamiltonian which describes the essential interactions in the helix at the level of the base pair and incorporates twisting and bending degrees of freedom. Helix unwinding and bubble formation patterns are consistently computed by a path integral method that sums over a large number of molecule configurations compatible with the model potential. The path ensembles are determined, as a function of temperature, by minimizing the free energy of the system. Fluctuational openings appear along the helix to release the stress due to the bending of the molecule backbone. In agreement with the experimental findings, base pair disruptions are found with larger probability in the smallest minicircle of 66 bps whose bending angle is ∼6°. For this minicircle, a sizeable untwisting is obtained with the helical repeat showing a step-like increase at T = 315 K. The method can be generalized to determine the bubble probability profiles of open ends linear sequences.

TfReg: calculating DNA and RNA melting temperatures and opening profiles with mesoscopic models

SUMMARY

The mesoscopic statistical physics models, known generically as Peyrard-Bishop (PB) models, have found many applications for the study of oligonucleotide properties. Unfortunately, PB models have not reached a wider non-specialized audience for the lack of freely available software implementations. Here we present an extensible C++ implementation of four variants of the PB model, which allows the user to calculate melting temperatures from tested model parameters. Even for a non-specialist, it should be straightforward to change these parameters to reflect different experimental environments or different types of oligonucleotides. For users with some proficiency in C++ programming, it should be feasible to extend the code to other PB models owing to the generic programming implementation adopted for TfReg. Pre-calculated parameters are included that allow the immediate calculation of melting temperatures and thermal equivalence indexes for DNA and RNA.

AVAILABILITY

C++ source code and compiled binaries for several Linux distributions are available from https://sites.google.com/site/geraldweberufmg/tfreg and from OpenSuse build service at http://build.opensuse.org.

CONTACT

gweberbh@gmail.com

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.