MOLECULAR MODELING OF HUMAN HAIR 473 glycine (19%), serine (22.4%), and lysine (5%). The UHSP by McKinnon et al. gives cysteine (36%), glycine (16%), serine (24%), and lysine (5%). These two UHSPs are similar and we could in principle use either sequence for our model. For this initial model we chose the one by Yahagi et al. (9) because it is the smaller of the two (156 vs 168 total amino acids). As described in detail in the Experimental section, a beta sheet model of polyalanine was fi rst constructed based on the conformation obtained from the Protein Data Bank. Next, the amino acid sequence was changed to fi t the amino acid sequence of the Yahagi UHSP (see Figure 1). The fully constructed beta sheet was then duplicated to several beta sheets that were stacked parallel to each other in the Y direction, as shown in Figure 2. The Yahagi UHSP sequence provides cysteine residues at the top of each fold, and in that manner it provides an anchor for 18-MEA at a distance along the chain of approximately 0.9 nm, with an intersheet spacing of about 0.7 nm, thus providing an inter-attachment distance along the UHSP of about 0.9 nm. This intersheet distance of 0.7 nm provides spacings relatively close to the calculation of 0.936 nm in a “pseudo-hexagonal array,” as described by Swift (14), and a beta sheet density of 1.42 gm/cm3compared with Allworden membrane fractions, as described by Allen et al. (6), that varied from 1.39 to 1.54 g/cm3. This outer layer of the epicuticle after relaxation and energy minimization (see Experi- mental section and Figure 1) was approximately 5-nm thick. The relaxation procedure allowed energy minimization while retaining cysteine residues on the top folds and main- tained the essential backbone of the beta sheets. The 18-MEA molecules were then at- tached, and after 120 ps of MD simulation followed by energy minimization, the average thickness of 18-MEA was found to be 1.08 ± 0.2 nm, in good agreement with the XPS results {1 ± 0.5 nm} of Ward et al. (2). This energy minimization procedure confi rms that 18-MEA chains bound by thioester linkages to a UHSP surface at attachment distances of approximately 0.9 will bend back on themselves to the thickness found by Ward et al. on Soxhlet-extracted wool fi ber, as suggested by Zahn et al. (4). The 18-MEA chains interact with each other by van der Waals attractive interactions. The 18-MEA chains are quite mobile, and without any other mol- ecule between them they can adopt different conformations to energetically favor inter- chain interactions within themselves as well as van der Waals interactions with hair surface amino acids at the top folds of the beta sheets of the epicuticle protein complex. The preliminary molecular model described in this paper (Figures 3–5) was selected and evaluated from eight different skeleton models, and it satisfi es the following requirements better than the other models: A 1.08 ± 0.2-nm thickness of 18-MEA at the surface after equilibration following MD simulations. The variance of this 18-MEA thickness is smaller than that of Ward et al. (2), but the actual thickness is consistent with their XPS measurements, and the attachment of 18-MEA to the epicuticle proteins by thioester linkages through the cysteine residues of a UHSP arranged in a beta sheet is consistent with the description by Negri et al. (1,15). This molecular model shows that the beta folds of this UHSP are perpendicular to the surface, as in Figures 1 and 2. Furthermore, this arrangement allows vacant spaces be- tween 18-MEA molecules, as shown in Figure 6B, for free lipid moieties consistent with our conclusions from XPS measurements by Carr et al. (16) and others (17,18) described in the next section of this paper.

JOURNAL OF COSMETIC SCIENCE 474 FREE LIPID IN THE SURFACE LAYERS Several different laboratories have analyzed the outer surface of wool and human hair via XPS, examining the outer 2 to 4 nanometers of the hair surface (2, 16–18). As indicated, Ward et al. (2) estimated the thickness of the lipid layer of 18-MEA at 1 ± 0.5 nm from carbon/nitrogen analysis, assuming XPS examines the top 3 nm. Carr et al. (16) estimated 60% protein and 40% lipid in the top 3 nm of Soxhlet-extracted wool fi ber. This estimate provides for 36% 18-MEA (at a 1.1-nm thickness) and 4% free lipid (non-covalently bound) in the top 3 nm of this wool sample. A related estimate using data from Robbins and Bahl (18) on “virgin” hair after shampooing with a sodium laureth-2 sulfate-containing shampoo provided 12% free lipid in the top 3 nm. Capablanca and Watt (19) examined wool fi ber that had been washed with detergent and extracted with various solvents using a streaming potential method to measure zeta po- tentials, from which they estimated the effect of extracted lipid on the isoelectric point of wool fi ber. These scientists found an appreciable effect of free lipid (solvent-extractable lipid) on the isoelectric point, with surfactant-scoured wool having an isoelectric point of 3.3, but after extraction of this same wool with the most effective lipid solvent they found an isoelectric point of 4.5. These data show that the true isoelectric point of the hair sur- face proteins is close to 4.5 and that free lipid, which contains fatty acids, is an important and essential component of the surface (top few nanometers) of animal hairs, especially hair or wool in good condition that has only been scoured or cleaned with surfactants or sham- poos. Furthermore, the more free lipid in these surface layers, the lower the isoelectric point of the fi bers, suggesting that free lipid is an integral part of the surface of animal hairs after and between normal shampooing and scouring treatments. We intend to explore the inclusion of free lipid in our keratin fi ber surface models with improved software to see if free lipid deposited between 18-MEA molecules affects the thickness of 18-MEA on the hair surface, i.e., if free lipid moieties between 18-MEA Figure 6. (A) Partially relaxed model to represent “bare-hair” (virgin hair) surface structures. Snapshots were taken after 40 ps of molecular dynamics simulations where top folds and 18-MEA were allowed to move during simulation. (B). Side view with van der Waals spheres for 18-MEA. Vacancies between 18-MEA res- idues can be seen, possibly available for free lipid deposits/interactions.