The DNA G-quadruplex (GQ) displays structural polymorphisms,
and
interactions between its loops and flanking sequences critically determine
which of the diverse GQ conformers is adopted. All-atom molecular
dynamics (MD) simulations of GQs are computationally challenging due
to slow folding times and force field (ff) artifacts. In an earlier
study, a direct folding simulation of the simplest DNA GQ (TBA15)
was first reported using a modified version of the AMBER bsc1 ff (bsc1_vdW
ff). Despite this successful folding simulation, it was later found
that the bsc1_vdW ff is somewhat limited in terms of describing loop
structures of GQs, which is problematic because GQ loop regions play
key roles in ligand binding to modulate GQ activities. In this study,
we further modified the bsc1_vdW ff to enhance the GQ loop prediction
by fine-tuning a limited number of van der Waals (vdW) parameters
of the standard AMBER bsc1 ff to improve the GQ loop distribution
of a target GQ system (three-layered antiparallel GQ; mHtel21). Test
simulations of this newly generated ff (bsc1_vdWL ff) on DNA GQs with
diverse topologies (hybrid1, hybrid2, and parallel propeller) revealed
that loop structures were predicted more accurately than by the bsc1_vdW
ff. We consider that enhanced sampling MD simulation methods in combination
with bsc1_vdWL provide useful simulation protocols for resolving outstanding
issues of DNA GQ folding and GQ/ligand binding at the all-atom level.