Abstract
Aquaporins (AQPs) are classical water channels that also conduct small gas molecules such as O2 and CO2 across the membrane. The hydrophobic central pore, located at the fourfold symmetry axis of an AQP tetrameric architecture, has been proposed to constitute the most optimal pathway for gas transport, although monomeric water pores can also contribute somewhat to permeation of less hydrophobic species. Here, we report a comparative molecular dynamics (MD) study of gas permeability in a plant AQP and a mammalian AQP1, taking advantage of complementary computational protocols including flooding simulations, umbrella sampling, and implicit ligand sampling. PIP2;1 AQPs, present in plants, are experimentally reported to have lower gas permeability than AQP1, which is present both in plants and animals. Using the spinach PIP2;1 (SoPIP2;1) and bovine AQP1 (bAQP1) as the models, the study unravels the specific structural features controlling the permeability of the central pore to gases. In SoPIP2;1, residue Trp79, which is highly conserved in the plant PIP2;1 family and lines directly the central pore, forms a major constriction region and the main barrier against gas permeation. Notably, the occluding conformation of the four Trp79 residues from the four monomers is stabilized by another conserved residue, Phe207 in the central pore. Sequence and structural comparisons show that both of these residues are replaced by less bulky residues in AQP1, for example, by Leu56 and Ala179, respectively, in bAQP1. The role of Phe207 residues in hindering gas permeation through SoPIP2;1 is confirmed by in silico alanine substitution, which reveals its effect on the local constriction produced by Trp79 residues. Conversely, by mutating Leu56 to tryptophan and Ala179 to phenylalanine in bAQP1, we engineer the protein to a less permeable gas channel.