What could be the molecular basis for the protein to twist and undergo a conformational change intrinsically (that does not involve any substrate binding event) ?
Yes. All objects try to minimize their surface area - helices are more compact than extended strands, supertwined helix bundles are more compact than separate helices lying alongside each other, likewise: beta-sheets will curl up into a barrel and that can twist more after it has formed.
Even smaller beta sheets twist: appart from issues of minimizing exposed area, beta-sheets are twisted due to the optimum geometry of their H-bonds, and it depends, among other things, on whether they're formed by parallel or anti-parallel beta-strands.
For a good discussion see the book by Finkelstein and Ptitsyn "Protein Physics".