Using molecular simulation to understand the skin barrier

Abstract

Skin's effectiveness as a barrier to permeation of water and other chemicals rests almost entirely in the outermost layer of the epidermis, the stratum corneum (SC), which consists of layers of corneocytes surrounded by highly organized lipid lamellae. As the only continuous path through the SC, transdermal permeation necessarily involves diffusion through these lipid layers. The role of the SC as a protective barrier is supported by its exceptional lipid composition consisting of ceramides (CERs), cholesterol (CHOL), and free fatty acids (FFAs) and the complete absence of phospholipids, which are present in most biological membranes. Molecular simulation, which provides molecular level detail of lipid configurations that can be connected with barrier function, has become a popular tool for studying SC lipid systems. We review this ever-increasing body of literature with the goals of (1) enabling the experimental skin community to understand, interpret and use the information generated from the simulations, (2) providing simulation experts with a solid background in the chemistry of SC lipids including the composition, structure and organization, and barrier function, and (3) presenting a state of the art picture of the field of SC lipid simulations, highlighting the difficulties and best practices for studying these systems, to encourage the generation of robust reproducible studies in the future. This review describes molecular simulation methodology and then critically examines results derived from simulations using atomistic and then coarse-grained models.

Keywords: Atomistic simulation; Coarse-grained models; Molecular dynamics simulation; Percutaneous permeability; Skin barrier function; Stratum corneum lipids.

Fig. 1.

Number of original research papers published since 2000 that describe simulations of SC lipids in the context of skin (i.e., a simulation that includes CHOL may not be included).

Fig. 2.

Structure and nomenclature for the 12 most prevalent CER subclasses found in human SC, designated as CER Z_FAZ_SB where Z_FA and Z_SB represent the one or two letter abbreviations for the fatty acid and sphingoid base, respectively. The complete structure for CER NS is presented as an example.

Fig. 3.

Schematic illustrating the organization and packing of the lamellar phases observed in the SC (redrawn from Pilgram et al. [67])). The phases are classified as orthorhombic or crystalline, hexagonal or gel, and liquid-crystalline or liquid-ordered. These terminologies are used interchangeably in the literature. Lipids in the liquid crystalline phase display lateral and rotational movements. In the hexagonal packing, hydrocarbon chains can rotate freely around their axes, whereas lipids in the orthorhombic packing are in solid state and packed more closely in one direction. *In this paper the word gel is also used in some contexts to describe a phase with limited mobility, which could be either orthorhombic or hexagonal.

Fig. 4.

Proposed two-dimensional molecular model arrangements of the unit cell for the LPP (a-c) and SPP (d-g) in the SC lipid matrix. In (a) and (c) ceramides 1, 2, 3, 4 and 6 correspond to CERs EOS, NS, NP, AS, and AP, respectively. (All figures have been reprinted with permission from the appropriate journal publisher. (f) is reprinted with permission from “Different Phase Behavior and Packing of Ceramides with Long (C16) and Very Long (C24) Acyls in Model Membranes: Infrared Spectroscopy Using Deuterated Lipids” by Školová B, et al., 2014, J Phys Chem B 118, p. 10468. Copyright 2014 from American Chemical Society.)

Fig. 5.

Schematic illustrations of (a) a hydrated bilayer (HBL); (b) a two hydrated bilayer (4-leaflet) stack (2-HBL), which includes water slabs in contact with headgroups of the two outer leaflets and intermembrane water between the bilayers; (c) a 4-leaflet hydrated multilayer stack (4-HML), which includes a water slab on the headgroups of the outer leaflets but no intermembrane water between the bilayers; (d) a 6-leaflet hydrated multilayer stack (6-HML); (e) a simulation box containing one hydrated bilayer (HBL) with periodic images above and below, and (f) a simulation box containing a 4-leaflet dehydrated multilayer stack (4-DML) with periodic images above and below. In (e) and (f) the simulation box is highlighted and the periodic images are faded.

Fig. 6.

Total (dotted line), lipid (solid line), and water (dashed line) mass density profiles for a pure CER NS C24 bilayer simulation from Moore et al. [115] (snapshot on top) showing the bilayer thickness calculated using the d_HH,m (48.5 Å, calculated from the lipid mass density profile), d_FWHM (57.3 Å) and d_W,1/e (55.0 Å) methods. Values obtained using methods that do not use density profiles are d_V = 56.0 Å, d_REF (O & N) = 51.9 Å, and d_REF (O) = 52.0 Å.

Fig. 7.

Comparisons of the S_CH, S_CC, mean tilt angle, and S₂ order parameters for the fatty acid chain of CER NS C24 in bilayers of either pure CER NS C24 or a 1:3 molar ratio with CER NS C16 derived from simulation results reported by Moore et al. [115]. Carbon number 24 corresponds to the terminal methyl with structure CH₃.

Fig. 8.

Schematic showing the MARTINI CG mapping and beads assigned for (a) CHOL in the original MARTINI force field [222], (b) CHOL in the 2015 update of the MARTINI force field where the dashed lines show the addition of virtual sites [237], (c) MARTINI water model illustrating the 4 to 1 water mapping, and (d) FFA C24 [222].

Fig. 9.

A summary of published CG MARTINI models for CERs: (a) CER NS C16 [145,155,212,221,240], (b) CER NP C24 [232], and (c) CER AP C24 [233]. The mapping scheme for each model is shown to the left of its CG representation. The CG representations denote the MARTINI bead type (inside each bead) as well as equilibrium bond lengths (between beads in nm) and bond angles (in degrees). CER NS C16 models are redrawn from Podewitz et al. [145].

Fig. 10.

Coarse-grained mapping for (a) FFA C16 in the IBI model [210], (b) CHOL in both IBI [209] and MS-IBI [110], and (c) CER NS C24 in the MS-IBI model [151]. The MS-IBI mapping for FFAs (not shown) is the same as in the IBI model except that tail beads are mapped 3:1, as in the fatty acid and sphingoid tails of the CER NS, rather than 4:1 [151].

Fig. 11.

Coarse-grained mapping for CHOL used in the SDK model [234].