Thickness,
area, and lipid tail order vs cholesterol concentration in POPC bilayers
compared between experiment, united-atom, and different coarse-grained
simulations. A: Bilayer thickness (phosphate-phosphate distance across
the leaflets), averaged over a grid as described in the Methods; experimental headgroup-headgroup distance (dHH) data are given from Hodzic et al.52,53 B: Difference in bilayer thickness from 0% cholesterol. C: Projected
area per molecule, with experimental areas in monolayers from Smaby
et al. at a surface pressure of 30 mN/m;56 D: estimated slope of the AM curve as
a function of mol % cholesterol; the slope between two consecutive Xcholxi and xi+1 is plotted at (xi + xi+1)/2. Panel E shows the C2B-D3B P2 order parameter for the oleoyl chain and panel F shows the C2A-C3A P2 order parameter for the palmitoyl chain. Experimental SCH data from Ferreira et al.43 (see Figure S8) are converted
to estimated P2 order parameters based
on a linear model relating SCH and P2 calculated from UA simulations, as described
in the Methods and Figure
S7. For all panels, black indicates the united-atom model.36 For coarse-grained simulations, the phospholipid
model is indicated by blue lines (MARTINI) or red lines (angle-corrected
POPC), while the sterol model is indicated by circles (MARTINI cholesterol)
or triangles (optimized cholesterol, that is, CT3-Me2b). Where available, experimental data are indicated in green.
Error bars indicate the standard deviation of each metric over simulation
time.