A critical review of computational efforts towards identifying secondary structure elements in polylactic acid (PLA)

Abstract

Polylactic acid (PLA) may be regarded as an analogue of a poly-alanine oligo/polypeptide, where the amino group has been replaced by a hydroxyl. As a consequence, a series of studies have explored the possibility that PLA can adopt peptide-type secondary structures – i.e., repetitive structural patterns characterized by intramolecular hydrogen bonds between neighboring functional groups. To this end, computational techniques (molecular mechanics, semiempirical, Hartree-Fock, density functional theory DFT) geometry optimizations of isolated oligomers of lactic acid (generally ten-unit oligomers), or oligomers attached to solid surfaces, or dimers have been reported, as well as spectral simulations thereof - looking at relative stabilities of helices (α, π, 310), and β sheets. A significant variation in the predicted structures and spectra was noted, depending on the computational method employed. With the most accurate method available (a DFT functional parametrized especially for describing non-covalent interactions), in isolated PLA models the π helix was found to be the most likely structure, closely followed by the 310 helix, and β sheets being the least stable. We review here these data and add two important elements: (1) first, a comparison with an experimentally-derived model of PLA, proposed by De Santis, and (2) second, a Ramachandran analysis of the Φ and Ψ angles in the optimized geometries. It is shown that (1) the De Santis structure is in fact slightly more stable than the helices, and (2) the optimized geometries in fact stray far from the initial Φ, Ψ values – to the extent that all of the peptide-like secondary structures in fact end up as turns (mostly type III β turns), while the DFT-optimized De Santis structure has no classical correspondent in the Ramachandran series of secondary structures.