Modeling Diffusion and Permeation Across the Stratum Corneum Lipid Barrier

TL;DR

This study uses atomistic MD to quantify diffusion and permeation of two skin-oil oxidation products, acetone and 6-MHO, plus water, across a stratum corneum lipid membrane modeled in the short periodicity phase. By combining umbrella sampling with two complementary estimators of position-dependent diffusivity—VACF and PACF—and validating with propagator analysis, the authors obtain upper and lower bounds on the true diffusivity profile $D(z)$ and propagate these to compute transmembrane permeabilities within the isodiffusivity framework. They find that permeation is dominated by energetic barriers ($ΔF(z)$) rather than by mobility ($D(z)$), with permeabilities spanning about an order of magnitude between the diffusion bounds, and FE-method differences (US vs. WTM) contributing far less to the final transport predictions. The results provide molecular-scale constraints for indoor air chemistry models concerned with skin-related VOC transport and establish a transferable framework for linking atomistic transport to macroscopic exposure models, while outlining future directions to incorporate more realistic SC architectures and robust diffusivity estimators.

Abstract

Human skin oils are a major sink for ozone in densely occupied indoor environments. Understanding how the resulting volatile and semivolatile organic oxidation products influence indoor air chemistry requires accurate representations not only of their emission into indoor air but also of their transport across the outermost skin barrier, the stratum corneum. Using molecular dynamics simulations, we investigate the passive permeation of acetone, 6-methyl-5-hepten-2-one, and water -- two representative products of skin-oil oxidation and a reference compound -- through a model stratum corneum lipid membrane. We determine position-dependent diffusivities using two complementary analyses based on the same set of simulations and evaluate their accuracy through a propagator analysis. The two approaches provide upper and lower bounds for the true diffusivity, which, when combined with previously reported free-energy profiles, yield permeabilities relevant for modeling macroscopic skin transport. Our results show that permeation is governed primarily by energetic barriers rather than by molecular mobility, and that the predicted transport coefficients vary by about one order of magnitude depending on the chosen diffusivity estimator. These findings provide molecular-level constraints for parameters used in indoor air chemistry models and establish a transferable framework for linking atomistic transport mechanisms to large-scale simulations of human exposure and indoor air quality.

Figures (8)

• Figure 1: Chemical structures of acetone and 6-MHO, and a representative snapshot of the model system. The atomistic lipid membrane is composed of cholesterol (orange), lignoceric acid (yellow), and ceramide NS (green). In this study, we investigate solute permeation across the membrane along the $z$-axis. Figure adapted with permission from ref. thomas_insights_2025.
• Figure 2: Transmembrane diffusivity profiles obtained from the velocity autocorrelation function (VACF) and position autocorrelation function (PACF) analyses. Quantitative and qualitative discrepancies are evident near the membrane center. Solid lines are shown to guide the eye, and dashed lines indicate reference values for the aqueous-phase bulk diffusivities. In the water profile, two representative data points—one in the bulk region (black) and one at the membrane center (red)—are highlighted for further analysis.
• Figure 3: Illustration of determining diffusivities from velocity autocorrelation functions. The quantity $D(s)$, computed from eq. \ref{['eq:ds']}, must be extrapolated to zero by fitting a straight line over a range of $s$ values where the function remains well-behaved. This procedure performs reliably in the bulk phase but is highly sensitive to the placement of the tangent line for windows at the membrane center. The values obtained using the automated extrapolation procedure of Rowley et al.gaalswyk_generalized_2016 are indicated by the black and red points in Figure \ref{['fig:diffusivities']}.
• Figure 4: Illustration of determining diffusivities from position autocorrelation functions (PACFs). Shown is the estimated diffusivity, $D(t)$, as a function of the PACF integration cutoff time in two selected regions of the membrane. In the absence of noise, $D(t)$ should reach a plateau. A clear plateau is visible in the bulk water region but not at the membrane center. The values corresponding to a fixed cutoff of $t = 10ps$ are indicated by the black and red points in Figure \ref{['fig:diffusivities']}.
• Figure 5: Free-energy profiles for the translocation of all solutes across the model stratum corneum membrane, obtained from umbrella sampling (US) and well-tempered metadynamics (WTM).thomas_insights_2025 The two methods produce consistent overall features, with only minor quantitative differences. Figure adapted with permission from ref. thomas_insights_2025.