The G-triplex DNA.

Nucleic acids represent the alphabet of the cellular language and through their sequence and topology regulate vital cellular functions. In recent years, it has been found that many variations from the Watson–Crick duplex structure play key roles in many cellular processes. Examples are hairpins, cruciforms, parallel-stranded duplexes, triplexes, G-quadruplexes, and the i-motif. These structures can be formed by nucleotide sequences distributed throughout the whole human genome, their location is not random and often associated with human diseases. These complexes are formed from one to four strands, stabilized by base stacking and hydrogen bond interactions, with a variety of non-standard pairings. For instance, DNA triplexes can present G:G-C, A:A-T, C:G-C, and T:A-T pairings, with two strands in the standard Watson–Crick duplex structure (i.e. G-C and A-T) and the third one lying in the major groove of the duplex. In contrast, G-quadruplexes are four-stranded structures stabilized by stacking of two or more guanine tetrads (Figure 1). These examples highlight the structural polymorphism of DNA and suggest that other structures might exist, perhaps with specific cellular functions that are, to date, unknown. Herein, using metadynamics simulations, we have identified a stable folding intermediate of the thrombin binding aptamer (TBA) quadruplex. This intermediate is characterized by a “G-triplex” structure, having G:G:G triad planes stabilized by an array of Hoogsteen-like hydrogen-bonds (Figure 1). This kind of structure has been already hypothesized in other investigations on different DNA sequences, but never experimentally proven. Herein, for the first time, we have structurally and thermodynamically characterized this DNA structural motif, through a combination of biophysical experiments. Well-tempered metadynamics simulations have been used to study the folding of TBA, which is a 15-mer oligonucleotide (5’-dGGTTGGTGTGGTTGG-3’) organized in an anti-parallel monomolecular G-quadruplex with a chairlike structure (Figure 2a). This structure consists of two Gtetrads, able to coordinate a metal ion at the center, connected by two TT loops and a single TGT loop. Metadynamics accelerates the sampling, adding a bias on a few degrees of freedom of the system, called collective variables (CVs). In such a way, long time scale events, such as ligand/protein docking or protein/DNA folding, can be sampled in an affordable computational time and the free energy surface (FES) of the process can be computed. In the present case, the FES was calculated as a function of two CVs, the radius of gyration CV defined by the oxygen atoms of the guanines forming the G-tetrads and a second CV that counts the number of hydrogen bonds between these guanines (see Supporting Information). Looking at the FES obtained after approximately 80 ns of metadynamics simulation, three main energy minima can be identified (Figure 2b). The deepest one, basin A, corresponds to the experimental G-quadruplex structure of TBA. In the second minimum, basin B, TBA shows a partial opening of the 3’ end with residue G15