The hypothesis of intrinsic backbone compression in B-DNA,that first emerged from free MD simulations of A-tract induced static bends,assumes that,under physiological conditions,the equilibrium specific length of the sugar-phosphate backbone,considered as a restrained polymer attached to a cylindrical surface,is slightly longer than necessary for optimal base stacking.Such a mismatch would cause a geometric frustration and can result in a much more complex behavior of the DNA structure than it is usually assumed.The backbone "pushes" stacked base pairs,forcing them to increase the helical twist and rise.The stacking interactions oppose this and,as a result,the backbone increases its length by deviating from its regular spiral trace on the cylindrical surface of the double helix,which causes quasi-sinusoidal modulations of the DNA grooves and local intrinsic curvature.This mechanism explains the origin of non-local sequence effects and offers new interpretations to some long-standing puzzles of intrinsically bent DNA.It also changes significantly the common view of DNA dynamics because,under certain conditions,the aforementioned geometric frustration may result in macroscopically slow relaxation processes.