Computer Folding of Parallel DNA G-Quadruplex: Hitchhiker's Guide to the Conformational Space

Abstract

Guanine quadruplexes (GQs) play crucial roles in various biological processes, and understanding their folding pathways provides insight into their stability, dynamics, and functions. This knowledge aids in designing therapeutic strategies, as GQs are potential targets for anticancer drugs and other therapeutics. Although experimental and theoretical techniques have provided valuable insights into different stages of the GQ folding, the structural complexity of GQs poses significant challenges, and our understanding remains incomplete. This study introduces a novel computational protocol for folding an entire GQ from single-strand conformation to its native state. By combining two complementary enhanced sampling techniques, we were able to model folding pathways, encompassing a diverse range of intermediates. Although our investigation of the GQ free energy surface (FES) is focused solely on the folding of the all-anti parallel GQ topology, this protocol has the potential to be adapted for the folding of systems with more complex folding landscapes.

Keywords: DNA quadruplex; computational folding; enhanced sampling; kinetic partitioning mechanism; metadynamics; molecular dynamics; nudged elastic band; pathCV; transition path sampling.

FIGURE 1

Possible folding pathway leading to the all‐anti parallel‐stranded GQ.

FIGURE 2

Workflow for preparing structures and pathCVs for calculation of free energy profiles of hairpin (7‐mer), triplex (11‐mer) and complete GQ (15‐mer) folding.

FIGURE 3

(A) Free energy profile of the parallel hairpin folding pathway, with error bars representing the standard error of the mean (SEM) indicated by red lines. The blue arrows denote free energy corrections for bias resulting from restraints on the orthogonal distance to pathCV (z coordinate). The dotted vertical line indicates the fully folded GQ hairpin state. (B) Representative structures along the free energy profile (guanines are shown in light and dark blue, adenines are in white, magenta lines correspond to hydrogen bonding, interacting Watson–Crick and Hoogsteen edges are highlighted in cyan and red, respectively).

FIGURE 4

Free energy surface (FES) of the 2D ST‐MetaD simulation of 11‐mer, highlighting four main minima (I‐IV) and two minimum energy paths (dashed violet lines) from single strand to G‐triplex. For each minimum, a structural scheme of a representative structure is shown, with guanines depicted in shades of blue, hydrogen bonds indicated by magenta dotted lines, and interacting edges are highlighted in cyan and red for Watson–Crick and Hoogsteen edges, respectively.

FIGURE 5

The free energy profiles (FEPs) of the folding of a parallel G‐triplex along pathway passing through G‐cross structure at (A) 5′‐side and (B) 3′‐side. The FEPs along minimum energy path (MEP) identified in the 2D ST‐MetaD (black line) is compared with the corresponding FEPs of optimized folding paths. The pathCV is normalized to facilitate comparison.

FIGURE 6

(A) The 3D cube representing six unique folding pathways from single strand to fully folded parallel GQ, each depicted in a different color. The folding states at the vertices are labeled based on the state of three consecutive G‐harpins, where “s” stands for single strand, “C” for G‐cross, and “H” for G‐hairpin. (B) The FEPs of the six pathways are shown in color scheme corresponding to paths on panel A. (C) Structural schemes of all folding structures at vertices of the cube.

FIGURE 7

Average number of cations with at least tetracoordination to the guanines (upper graph) and the average number of native hydrogen bonds (lower graph) along the GQ folding pathway. Representative 3D structures along the folding pathways with ion binding sites highlighted by spots of high ion density are shown.