Evidence of Free Fatty Acid Interdigitation in Stratum Corneum Model Membranes Based on Ceramide [AP] by Deuterium Labelling

Biochimica et Biophysica Acta 1788 (2009) 2194–2203

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbamem

Evidence of free fatty acid interdigitation in stratum corneum model membranes based on ceramide [AP] by deuterium labelling

Annett Schroeter a, Mikhail A. Kiselev b, Thomas Hauß c,d, Silva Dante c,1, Reinhard H.H. Neubert a,⁎ a Martin-Luther-Universität Halle-Wittenberg, Institute of Pharmacy, Department of Pharmaceutics and Biopharmaceutics, Halle (Saale), Germany b Frank Laboratory of Neutron Physics, Joint Institute for Nuclear Research, Dubna, Moscow Reg., Russia c Helmholtz-Zentrum Berlin für Materialien und Energie, Berlin, Germany d Technische Universität Darmstadt, Dept. of Chemistry, Physical Biochemistry, Darmstadt, Germany article info abstract

Article history: This research paper provides direct evidence concerning the localisation of free fatty acids in stratum Received 5 September 2008 corneum lipid model membranes. We employed partially deuterated free fatty acids to gain further Received in revised form 24 July 2009 information about the assembly of a stratum corneum lipid model membrane based on a ceramide of the Accepted 28 July 2009 phytosphingosine-type (ceramide [AP]) with particular respect to the position of the deuterated groups of Available online 6 August 2009 the free fatty acids. The application of behenic-22,22,22-d3-acid and cerotic-12,12,13,13-d4-acid conﬁrmed that the short-chain ceramide [AP] forces the longer-chained free fatty acids to incorporate into the bilayer Keywords: Ceramide created by ceramide [AP]. The ceramide [AP] molecules determine the structural assembly of this model Stratum corneum membrane and obligate the long-chain free fatty acids to either arrange inside this formation or to separate Lipid model membrane as a fatty acid rich phase. Deuterated free fatty acid © 2009 Elsevier B.V. All rights reserved. Neutron diffraction

1. Introduction (CHOL) and its derivatives as well as free fatty acids (FFAs) [6–9].Itis generally accepted that the ceramides as the main constituents [10,11] The outermost layer of the mammalian skin, the stratum corneum play a key role in structuring and maintenance of the barrier function (SC), has been the object of various research papers as it exhibits the of the skin [12,13]. Up to now nine different types of ceramides have main skin barrier [1]. This superficial layer is made up of corneocytes been identified [14], of which the short-chain ceramides of the which are filled with the structural protein keratin. The corneocytes sphingosine, hydroxysphingosine and phytosphingosine-type (such are further embedded in a matrix of highly ordered lipids [2]. The lipid as CER[AP]) make up the largest amount quantitatively [14,15].In part of this matrix has been identified as the major barrier for contrast to most biological membranes the lipid part of the SC does not substances to penetrate through the SC [1,3–5]. contain phospholipids which further characterises the unique lipid Furthermore, the lipid matrix has been identified as the major composition of the SC lipid matrix. Taken all together, the SC lipid diffusion-rate limiting pathway [4,5]. Therefore, the comprehension matrix has a very complex composition, therefore it is difficult to of the nanostructure as well as the relative properties of the SC lipid assess the function of each lipid species for the barrier properties of the matrix on a molecular level is crucial for understanding of the SC. Consequently, it is necessary to use simplified model systems to penetration of dermal drugs through the skin. clarify the role of each individual lipid species within the lipid matrix It was further established that the SC lipid matrix is comprised of and to receive a detailed picture of the molecular organisation of the three major lipid classes, namely the ceramides (CER), cholesterol lipids in the stratum corneum. The application of a realistic, but simplified model membrane enables a systematic study of the impact of different lipids on the assembly and can give insights into the Abbreviations: CER[AP], N-(α-hydroxyoctadecanoyl)-phytosphingosine; CHOL, properties of the internal nanostructure of the lipid bilayers [16].In cholesterol; ChS, cholesterol sulphate; FFA, free fatty acid; d22BA, behenic-22,22,22- that sense the application of well-defined synthetic stratum corneum d3-acid; d12CA, cerotic-12,12,13,13-d4-acid; RH, relative humidity; SC, stratum – – corneum; SLD, scattering length density; LPP, long-periodicity phase; SPP, short- lipids [16 19] in contrast to isolated native human SC lipids [20 27] is periodicity phase the more favourable way to obtain detailed information as the natural ⁎ Corresponding author. Institute of Pharmacy, Department of Pharmaceutics and variability of the native lipids can be overcome. Biopharmaceutics, Martin-Luther-University, Wolfgang-Langenbeck-Str. 4, 06120 Halle In the presented study the SC lipid model membranes are based on (Saale), Germany. Tel.: +49 345 5525001; fax: +49 345 5527292. the very polar short-chain phytosphingosine-type ceramide [AP]. E-mail address: [email protected] (R.H.H. Neubert). 1 Present address: Italian Institute of Technology, Facility of Nanobiotechnology, Although the amount of CER[AP] found in the SC is in the range of Genove, Italy. only 4–7% [13,28] it is of interest to understand the function of this very

0005-2736/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bbamem.2009.07.024 A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203 2195 polar and short-chain ceramide. As reported recently [29] the alteration lipid model systems. The deuterated part of the lipid shows a distinct of the FFA chain length in membranes based on CER[AP] did not lead to increase in the neutron SLD profile. When compared to the membrane an increase in the repeat distance because the influence of the polarity of containing the protonated lipid the distribution of the deuterium atoms the head groups of CER[AP] outweighs the influence of the length of the can be calculated. Therefore, precise information about the location of hydrocarbon chain of the free fatty acids. deuterated labels within the lipid membrane is available and will be For the investigation of the internal nanostructure of stratum presented further on. corneum lipid matrices and its properties it could be demonstrated that X-ray and neutron scattering techniques are very powerful tools 2. Materials and methods [20,22–26,30–36]. The application of neutron diffraction has the distinct advantage as it offers the possibility of the determination of the signs of 2.1. Materials the structure factors by contrast variation. This contrast variation is achieved by partial exchange of water by heavy water [37]. The neutron The SC lipids used for the preparation of the model membranes are scattering length density (SLD) profiles calculated using the structure presented in Fig. 1. CER[AP] with a chain length of 18 C-atoms was factors can give information about the nanostructure of the lipid kindly donated by Evonik Goldschmidt GmbH (Essen, Germany), membranes. The evaluation of specific membrane parameters from the while cholesterol and cholesterol sulphate were acquired from Sigma- neutron SLD profiles such as the region of the polar headgroups, the Aldrich (Taufkirchen, Germany). The partially deuterated FFA doc- hydrophilic and lipophilic membrane section and the thickness of the osanoic-22,22,22-d -acid (further referred to as d behenic acid, intermembrane space [16] can further improve the knowledge and 3 22 d BA, 99.2 atom% D) and hexacosanoic-12,12,13,13-d -acid (further comprehension about SC lipid model membranes. In contrast, the 22 4 referred to as d cerotic acid, d CA, 99.2 atom% D) were purchased structural analysis in the X-ray experiment is not as easily accessible due 12 12 from Dr. Ehrenstorfer GmbH (Augsburg, Germany). to the complexity of the evaluation of the signs for the structure factors [38]. In addition to the advantages of the contrast variation the neutron 2.2. Preparation of multilamellar lipid membranes diffraction experiment allows the exact determination of a selectively labelled lipid species within the lipid membrane. The exchange of SC lipid model membranes were prepared in two different com- hydrogen by deuterium atoms enables this identification of the lipid positions, where only the deuterated FFA component was exchanged as species as the coherent scattering length of hydrogen and deuterium shown in Table 1. atoms differs significantly (hydrogen bH =−3.74 fm, deuterium bD = In order to produce a lipid mixture with a total lipid concentration of +6.67 fm). In X-ray diffraction experiments such an exact determina- 10 mg/ml the appropriate quantity of each lipid species was dissolved tion of the position is possible when the desired lipid is selectively in a chloroform:methanol mixture (2:1). To create multilamellar SC labelled with heavy atoms such as bromide or iodine, but this labelling lipid membranes a volume of 1200 µl of this mixture was spread over method has the drawback for the structural analysis of lipid membranes the surface of a quartz slide with the dimensions of 6.5 cm×2.5 cm as the insertion of such alien atoms may affect the structural assembly. according to the procedure by Seul and Sammon [39]. After the Therefore, the application of neutron diffraction as an equal supplement deposition the solvent was removed by evaporation firstly at atmo- to X-ray scattering can extend the knowledge of well-oriented multi- spheric pressure then under vacuum. Afterwards a subsequent heating lamellar lipid membranes and especially the experimental pool in the (at 75 °C) and cooling cycle at 100% relative humidity (RH) was applied. research of SC lipids. This annealing procedure was necessary in order to acquire an improved In our preceding work [29], which was focused on the influence of orientation of the membrane and thereby decreasing the mosaicity of the free fatty acid (FFA) chain length on the nanostructure of SC lipid the sample and to create one-dimensional crystal domains with model membranes based on ceramide [AP] (CER[AP]) we concluded approximately 1600 membrane layers and a mosaicity about 0.1°. that the methylen and methyl groups of the long-chain FFAs need to Such a preparation procedure is commonly used in neutron diffraction protrude into the opposing leaflet in order to arrange inside the bilayer experiments [40–42]. The improvement in the mosaicity due to this created by CER[AP]. The FFAs, therefore, need to interdigitate into the annealing process is demonstrated in Fig. 2, whereby Fig. 2A shows the membrane size determined by CER[AP]. It was demonstrated that CER model system containing deuterated behenic acid before the annealing [AP] creates a superstable structure, which is not influenced by the was performed, while Fig. 2B displays the rocking curve of the same alteration of the FFA chain length. The purpose of the presented work is sample after the heating and cooling cycle was applied. This procedure to portray the direct experimental evidence of this interdigitation improved the signal-to-noise ratio in the neutron diffraction experi- behaviour of the fatty acids by applying selectively deuterated behenic ment and was essential for the further analysis of the received data. The and cerotic acid to visualise the location of those free fatty acids in the SC thickness of the lipid film on the quartz slide was approximately 7.5 µm.

Fig. 1. Chemical structures of the constituents of the SC lipid model membranes under investigation. The nomenclature of the ceramides is according to [61], whereby the letter A means α-hydroxy fatty acid which is bound to the amide, while the second letter stands for the type of base of the ceramide (P … phytosphingosine). 2196 A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203

Table 1 two-dimensional position sensitive 3He detector with a sensitive area Composition of the quaternary system based on CER[AP]. of 19 cm×19 cm and a pixel size of 1.5 mm×1.5 mm was used. The Mixture Free fatty acid (FFA) Component ratio (w/w) distance from sample to detector amounts to 102.4 cm. The neutron (CER[AP]/CHOL/FFA/ChS) diffraction in the reﬂection setup was used to collect the data of the QuatBA Behenic acid, BA 57/24/9.5/9.5 one-dimension diffraction experiment.

Quatd22-BA d22 behenic acid, d22BA 55/25/15/5 All samples under investigation were measured in thermostated QuatCA Cerotic acid, CA 55/25/15/5 aluminium cans at a ﬁxed temperature of 20 °C and relative humidity Quatd12-CA d12 cerotic acid, d12CA 55/25/15/5 (RH) of 57%. An aqueous solution of sodium bromide was used to control The model membrane was always composed of CER[AP]/Cholesterol (CHOL)/FFA/ the humidity inside the measurement chamber. Details can be found Cholesterol sulphate (ChS). The FFA used was either deuterated or non-deuterated. elsewhere [43]. The variation of the difference of the scattering length density in the chamber (neutron contrast) was achieved by adjusting

the chamber atmosphere up to four different H2O/D2O compositions (92/8, 80/20, 50/50 and 0/100, w/w) in order to facilitate the assign- 2.3. Neutron diffraction measurements and data treatment ment of the phases [44]. The sample was equilibrated for 12h after each change of aqueous solution. The neutron diffraction measurements were carried out using the The diffraction intensities of each sample were recorded either as membrane diffractometer V1 at the Berlin Neutron Scattering Centre θ−2θ-scan (high mosaicity samples), or as rocking scan (ω-scan), of the Helmholtz Centre Berlin for Materials and Energy (Berlin, whereby the sample was rocked around the expected Bragg position Germany) with a cold source and a neutron wavelength λ of 5.23 Å. A θ,byθ±2°. Both ways allowed the collection of up to ﬁve orders of diffraction. Conventionally, the structure of the bilayers is analysed by con-

struction of the neutron scattering length density (SLD) proﬁles ρS(x)in a direction normal to the membrane

hmax ⋅π⋅ ⋅ ρ ð Þ 2 ∑ 2 h x ; ð Þ s x = a + b Fh cos 1 d h =1 d

where Fh is the structure factor of the diffraction peak of the order h and d is the lamellar repeat distance calculated from the position of the peak centre (PeakFit SYSTAT Software Inc., San Jose, USA). The terms a and b in Eq. (1) are free coefficients andpffiffiffiffiffiffiffi are unknown. The absolute value of the structure factors jFh j = hIh is given by the integrated intensity Ih (obtained with Gaussian fits to the experimental Bragg reflection) of the hth diffraction peak. Details of the evaluation of the neutron diffraction data are described elsewhere [37,44,45]. The localisation of the specific deuterium label within the lipid membrane is reflected by the positive difference in the SLD profile between samples containing the deuterated and protonated FFA. According to Eq. (2) the difference was calculated for each structure factor

Δ Deut ð Þ ð Þ: ð Þ Fh = Fh deut Fh prot 2

The difference SLD proﬁle can be calculated by combining Eqs. (1) and (2) to Eq. (3) [46–48].

hmax ⋅π⋅ ⋅ ΔρDeutð Þ 2 ∑ Δ Deut ⋅ 2 h x : ð Þ s x = Fh cos 3 d h =1 d

The lipid membrane structure is centrosymmetric, therefore it is possible to ﬁt the maxima resulting from the deuterated groups of the FFA with two identical Gaussian functions using Eq. (4), thereby receiving the exact position of the centre of those groups containing the deuterated label.

A 1 x x 2 1 x x 2 ρsymðxÞ = pffiffiffiffiffiffiffiffiffiffi exp 0 + exp 0 : ð4Þ 2πσ 2 σ 2 σ Fig. 2. Rocking curve around a fixed angle 2θ to verify the mosaicity of the sample composed of CER[AP]/CHOL/d CA/ChS. (A) Example for a sample with high mosaicity 12 ρsym before the annealing procedure was performed. (B) Illustration of the rocking curve of Therefore, the function (x) corresponds to the deuterium dis- the same sample after heating and cooling at 100% RH. tribution across the membrane. A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203 2197

3. Results and discussion Table 2 Structure factors (SF) and associated signs for the small phase of the sample containing either deuterated or protonated behenic acid in the composition CER[AP]/CHOL/d BA 3.1. Deuterated terminal methyl group of behenic acid 22 or BA/ChS (55/25/15/5) at 8% D2O and 57% RH.

− 1 − 1 In this approach the FFA behenic acid (BA) in the SC model mem- Diffraction order hQ,Å SF for CER[AP]/ Q,Å SF for CER[AP]/ brane composed of CER[AP]/CHOL/BA/ChS was deuterium labelled at CHOL/BA/ChS CHOL/d22BA/ChS the terminal methyl group in order to identify the exact position of 1 0.133 −8.96±0.05 0.131 −3.98±0.05 the label within this lipid model membrane and to prove that the 2 0.271 +4.32±0.09 0.266 +6.59±0.07 3 0.409 −3.59±0.11 0.407 −3.72±0.09 FFA protrudes into the adjacent layer [29]. As such model membranes 4 0.539 +1.02±0.09 0.522 +3.69±0.10 show a tendency towards phase separation, the application of partially deuterated FFA can also give evidence that the separated phase is an FFA-rich phase as was concluded in [29] by determining the (relative) concentration of deuterium in the two phases, respectively. for the main phase. The neutron scattering length density (SLD) proﬁles

The quality of the sample allowed the collection of five diffraction ρs(x) for both existent phases, namely phase1_small and phase2_main, orders as rocking curves. Fig. 3 represents the diffraction pattern of the were calculated separately according to Eq. (1) in order to compare the SC model membrane containing deuterated behenic acid after the final deuterated with the protonated membrane [45]. Fig. 4 presents the background correction was realised with the program PeakFit. The neutron SLD profile of phase1_small containing either protonated diffraction pattern of the investigated membrane displayed a two-phase behenic acid (BA) or the deuterated species (d22BA) where the system, whereby the scattering intensity of the phases differed deuteration is located at the terminal methyl group. In the neutron significantly. Consequently, the notation of the phases was made SLD profile the midplane of the lipid bilayer is centred at zero, while the according to the scattering intensity and position of the first diffraction hydrophilic head groups are positioned at the outer edges, contributing order: phase1_small (due to the smaller Q-value and the smaller the maxima at the edges of the neutron SLD profile [16].TheSLDprofile scattering intensity) and phase2_main because it exhibits the main ofthedeuteratedsampleclearlyreflects a distinct maximum at the scattering intensity. From the positions of a series of equidistant peaks centre of the membrane, which corresponds to the position of the

(Qn), the lamellar repeat unit d,ord-spacing of each lamellar phase was deuterated methyl groups of d22BA as can be seen in Fig. 4. Deut calculated using the equation Qn =2nπ/d,wherebyn stands for the The difference SLD profile Δρs (x) was calculated by applying order number of the diffraction peak. The lamellar repeat distance Eqs. (2) and (3) to exactly localise the position of the deuterated methyl amounts to d1 =47.4±0.8 Å for phase1_small and dmain=44.1±0.2Å groups. for the phase with the main scattering intensity, respectively. For As can be seen in Fig. 4 the thickness of the non-deuterated and the orientation the diffraction orders 1 to 4 of phase1_small as well as deuterated phases differ (dQuatBA =45.5Å, dQuat22BA =47.4 Å). This does diffraction orders 1 to 5 of the main phase has been indicated in Fig. 3. not affect the calculation of the difference SLD profile. Such an SLD Deut Small amounts of phase-separated cholesterol crystals can be identified difference profile Δρs (x) of the phase1_small of the model membrane from the [010] reflection which is located at Q=0.18Å−1, representing containing behenic acid is presented in Fig. 4.Themaximum diffraction from the triclinic crystal with the lattice parameters corresponds to the deuterium atoms, which are situated in the central a=14.172 Å, b=34.209 Å, c=10.481Å, and α=96.64°, β=90.67°, part of the bilayer, indicating that the deuterated methyl group of BA is γ=96.32° [49]. The crystalline CHOL does not influence the multi- positioned at the centre. The position was fitted by a Gaussian function – lamellar lipid organisation as described in other studies [19,50 52] and and the centre was determined at xC22D3 =0.01±0.04 Å. The fact, that has no effect for the further reconstruction of the Fourier analysis. the deuterated behenic acid could be identified to a high extent within From the integrated peak intensity the structure factors (SF) were the smaller phase shows that this phase is made up not solely, but to the evaluated and are listed in Table 2 for the smaller phase and in Table 3 a high extent of FFA and gives direct evidence of the existence of the so-

Fig. 3. Diffraction pattern of the sample containing deuterated behenic acid (CER[AP]/CHOL/d22BA/ChS (55/25/15/5)) measured as rocking scans at 5 different detector angles (6.8,

13.6, 20.4, 27.2, and 34.0), at 57% RH, 50% D2O and 20 °C. Up to 5 orders of diffractions are visible. For phase 1_small diffraction orders 1 to 4 and diffraction orders 1 to 5 for phase2_main (b) have been indicated for better understanding. The abbreviation CHOL indicates the [010] diffraction peak from phase-separated cholesterol crystals. 2198 A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203

Table 3 difference of the repeat distances between the hypothetical BA bilayer Structure factors (SF) and associated signs for the main phase of the sample containing and phase2_main amounts to approx. 10.9 Å, as the d-value for the main either deuterated or protonated behenic acid in the composed of CER[AP]/CHOL/d22BA phase was calculated to be dmain=44.1±0.2 Å. A compensation of this or BA/ChS (55/25/15/5) at 8% D2O and 57% RH. difference can be accomplished by the interdigitation of the terminal − 1 − 1 Diffraction order hQ,Å SF for CER[AP]/ Q,Å SF for CER[AP]/ heptyl (9 Å) or octyl (10.25 Å) residues of the behenic acid molecules. CHOL/BA/ChS CHOL/d22BA/ChS 1 0.142 −21.08±0.04 0.142 −18.05±0.05 2 0.283 +5.75±0.07 0.281 +6.65±0.08 3.2. Deuteration of the methylen groups 3 0.426 −6.64±0.09 0.427 −6.82±0.13 4 0.569 +3.03±0.09 0.571 +2.57±0.10 The longest free fatty acid investigated in this study was cerotic acid 5 0.710 −3.49±0.14 0.711 −3.06±0.12 (CA) speciﬁcally deuterated at the C-atoms C12 and C13 in order to further prove that the long-chain free fatty acids need to interdigitate into the membrane size dictated by the short-chain ceramide [AP]. The diffraction pattern of the sample containing the deuterated CA, called FFA-rich phase. Furthermore, to underline this experimental recorded as θ−2θ scan is presented in Fig. 6. As demonstrated for the result the following model should be applied. membrane containing behenic acid, the membranes containing CA also The estimated repeat distance d of a bilayer created by two adjacent exhibit a two-phase system at 20 °C and 57% relative humidity. For behenic acid molecules amounts to approximately 55 Å, assessed by the clarity, the diffraction peaks of both existent phases have been indicated values of 1.5 Å for a CH3-group and 1.25 Å for a CH2-group [53].Asthe in Fig. 6. In our preceding work [29] we concluded that the separated repeat distance d for phase1_small amounts to d1 =47.4±0.8 Å a phase with the smaller Q-value and scattering intensity is a so-called difference of approximately 7.6 Å occurs to the theoretical BA mem- FFA-rich phase, as the solubility of the long-chain FFAs within the phase brane. Therefore, this difference can be compensated by the interdig- created by CER[AP] is decreased. Furthermore, it was shown that the itation of the terminal butyl (5.25 Å) or pentyl (6.5 Å) residues of the nanostructure along with its structural parameters such as the thickness behenic acid chain. This minor interdigitation of the terminal butyl of the hydrophilic and hydrophobic layers of the main phase is not residue accounts for the broadness of the maximum in the difference SLD affected when the FFA chain length is increased [29].Themembrane proﬁle as seen Fig. 4. repeat distance d was determined for each phase and amounts for the

The neutron SLD profile of the second phase with the main scattering FFA-rich phase to d1 =56.3 Å ± 0.6 Å and to dmain=43.8Å±0.4Å for intensity (phase2_main) is displayed in Fig. 5.Thefigure further dem- the main phase. The neutron SLD profile of the FFA-rich (CA-rich) phase onstrates the comparison of the neutron SLD profiles of the protonated containing the specifically deuterated CA showed two distinct maxima and deuterated membranes. Additionally the difference SLD profile (see Fig. 7A). The corresponding structure factors for the CA-rich phase Deut Deut Δρs (x) of the protonated and deuterated phase2_main is included. are presented in Table 4. The difference SLD profile Δρs (x)ofthe Here, two overlapping maxima are visible, which resemble the protonated and the deuterated FFA-rich phases together with the distribution of the deuterium atoms. Therefore, two Gaussian functions corresponding fit of the difference profile by two Gaussian functions are Deut Deut according to Eq. (4) were used to fit the difference SLD profile Δρs (x) presented in Fig. 7B. The results of the fitting procedure of Δρs (x)by in order to identify the position of the label. The centre of each Gaussian two Gaussian functions ρsym(x) according to Eq. (4) are summarised in − function was determined at xC22D3 = 4.08± 0.02 Å and 4.08±0.02 Å, Table 6 and give the exact position of the CD2-groups within the CA-rich respectively. This result shows that the terminal methyl group of the phase. The CD2-groups of the deuterated cerotic acid d12CA are situated specifically deuterated behenic acid d BA is not positioned at the centre − 22 inside this phase at xC12D2C13D2 =14.05±0.08 Å and 14.05±0.08 Å, of the membrane as demonstrated for phase1_small of this membrane. respectively. As it was concluded that this separate phase is an FFA-rich This is due to the chain interdigitation of the FFA in this phase. Again, the phase the determination of the deuterated label of CA directly proves theoretical BA bilayer can be used to confirm this result. In this case a that this phase contains phase-separated CA.

Fig. 4. Behenic acid, small phase. neutron SLD profiles at 8% D2O of the membrane containing either deuterated d22BA (dashed line) or protonated BA (solid line). Dotted lines: corresponding errors. Long dash: difference SLD profile. Fat solid line: fit of the difference SLD profile by a Gaussian function (deuterium distribution). A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203 2199

Fig. 5. Behenic acid, main phase. Neutron SLD profiles at 8% D2O of the membrane containing either deuterated d22BA (dashed line) or protonated BA (solid line). Dotted lines: corresponding errors. Long dash: difference SLD profile. Fat solid line: fit of the difference SLD profile by two Gaussian functions (deuterium distribution).

Once more the theoretical FFA membrane model can be employed. Related to the fitted position at 14.05 Å, this would be inside these error Here, two opposing CA-molecules have an estimated repeat distance d margins. of approximately 65.5 Å. The difference to the d1-value of the FFA-rich In the case of the main phase of the SC model membrane containing Deut phase of the SC model membrane amounts to approx. 9.2 Å, which can deuterated CA the calculation of the difference SLD Δρs (x)resultedin be compensated by the interdigitation of the terminal heptyl residues of two broad maxima overlapping each other, as shown in Fig. 8.Thecorre- the CA chains. From the full width at half height (FWHH) it can be sponding structure factors are listed in Table 5. As the deuterated CD2- deduced that the CH2-groups region represents the largest region inside groups of the CA are detectable in this model membrane based on CER[AP] Deut the membrane [16]. Assuming that the relevant maxima of the Δρs (x) it can be stated that some amounts of this long-chain FFA incorporates into sym correspond to the region of C12D2-C13D2 with the centres x at 14.05 Å this phase. The fitting of two Gaussian functions ρ (x) of the difference Deut and −14.05 Å, respectively, the FWHH (7.37 Å) comprises the region SLD profile Δρs (x) revealed that the CA protrudes through the adjacent with the highest density of deuterium atoms. This region can be layer in order to fit into the membrane size created by CER[AP] (see Fig. 8). calculated as 14.05±1/2 FWHH (±10.36 Å to ±17.74 Å). In this The results received from the fitting are summarised in Table 6. hypothetical cerotic acid bilayer, the position of the label at C12 and C13 Applying the theoretical CA bilayer model yet again as described − can be estimated as xC12D2-C13D2 =+16.5Å and 16.5 Å, respectively. above a difference to the dmain-value of approx. 21.7 Å occurs, which

Fig. 6. Neutron diffraction pattern of the sample containing the deuterated cerotic acid (CER[AP]/CHOL/d12CA/ChS (55/25/15/5)) measured as θ−2θ-scan at 57% RH, with a H2O/

D2O ratio of 92/8 and T=20 °C. Up to 5 orders of diffraction peaks were detectable. For better understanding the two phases of the system have been indicated: L1a–L5a indicate the 5 diffraction orders of the FFA-rich phase, while L1b–L5b indicate the 5 diffraction orders for phase2_main, respectively. 2200 A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203

Fig. 7. Neutron scattering length density profile of the FFA-rich phase of the sample containing cerotic acid. (A) Comparison of the neutron SLD profile of the sample containing the deuterated CA (long dash) with the model membrane containing the protonated FA (solid line). (B) Cerotic acid, FFA-rich phase. Difference SLD profile for CA specifically deuterated at positions 12, 12, 13, 13 (–––). Dotted lines: corresponding errors. Fat solid line: fit of the difference SLD profile by two Gaussian functions (deuterium distribution).

can only be compensated by the interdigitation of the last 15 to 16 CH2 groups and the terminal methyl group of the CA chain. Table 4 Comparison of the structure factors (SF) of the CA-rich phase of the SC lipid membrane containing either deuterated or protonated cerotic acid (CER[AP]/CHOL/d12CA or CA/ 4. Conclusion

ChS, 55/25/15/5) at 8% D2O and 57% RH.

Diffraction order hQ,Å− 1 SF for CER[AP]/ Q,Å− 1 SF for CER[AP]/ The results presented in Figs. 5 and 8 give direct evidence of the

CHOL/CA/ChS CHOL/d12CA/ChS interdigitation of the long-chain FFAs in the membranes based on the 1 0.112 −5.66±0.06 0.115 −6.53±0.06 CER[AP]. It was shown in our previous studies that CER[AP] creates a 2 0.223 +2.05±0.06 0.229 −4.30±0.06 superstable membrane, in which the nanostructure is inﬂuenced 3 0.332 −3.79±0.09 0.343 −2.42±0.09 neither by the long-chain FFA [29], nor by long-chain ceramides such 4 0.444 +2.70±0.12 0.446 +6.22±0.12 as CER[EOS] [54]. Moreover, this nanostructure is stable under full 5 0.555 −3.66±0.15 0.564 −4.17±0.15 hydration [55]. This property of CER[AP]-based membranes refers to the A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203 2201

Fig. 8. Cerotic acid, main phase. Difference SLD profile of CA specifically deuterated at positions 12, 12, 13, 13 (- - -). Dotted lines: corresponding errors. Fat solid line: fit of the difference SLD profile by two Gaussian functions (deuterium distribution).

armature reinforcement model of the SC membrane structure [55], which and the polar head group is located in the centre, while the hydrocarbon in terms of the so-called brick-and-mortar-model could explain the tails are parallel in the hairpin conformation. mechanical properties of the SC lipid matrix of SC. The armature The main principals of the armature reinforcement model are reinforcement model is a further development of the single gel phase presented in Fig. 9. The narrow contact of two adjacent bilayer leaflets model [56] on the level of the bilayer nanostructure and was postulated is created by the FE conformation of the ceramide molecules as shown by Kiselev et al. [55] in order to explain results received in different and in Fig. 9A. Therefore, the ceramide in FE conformation acts as a bridge independent experiments: (i) SAXS investigations of hydrated SC or anchor between the adjacent leaflets. The formation of this bridge extracted from human skin showed that at 60% hydration (w/w) a is only possible due to the low intermembrane hydration and when shift to a smaller Q-value of the peak position indicating the 6.4 nm the lipids are in the gel phase to allow the interdigitation of the chains phase and the shoulder indicative of the unit cell with the repeat [60]. Further, this anchor prevents the lipid bilayers from moving distance of 13.4 nm disappeared. It was concluded that this unit cell is apart during the course of hydration. After hydration in water excess transformed to disordered lipid structure [57];(ii)whenSCisolated the intermembrane space increases as depicted in Fig. 9B as a result of from porcine skin is hydrated via water vapour (100% RH) or excess the transformation of the ceramide molecules from fully extended to water the characteristic lamellar repeat distance increases from 5.7 nm hairpin conformation, the so-called chain-flip transition. This space is to 6.2 nm, respectively, and increases further to 7.0 nm after longer then hydrated by water and the narrow contact between the layers excess hydration [58]. After further extensive hydration in excess water disappears. the first order diffraction peak disappeared, which is in good agreement Thus, the presented experimental results are important as a further with the decrease of the intensity of the first order diffraction peak from support of the armature reinforcement model of the lipid SC matrix the SC lipid model membrane observed by Kiselev et al. [16]. caused by the short-chain CER[AP] molecules. This new theoretical model is based on the existence of the short- chain ceramides in two conformations, the fully extended and the 5. Summary hairpin conformation [59] and the phenomena of the chain-flip transition of the CER[AP] molecules from fully extended to the hairpin The application of the specifically deuterated free fatty acids cerotic conformation [16,55]. When the ceramides are in fully extended and behenic acids supplied the direct experimental evidence that long- conformation (FE) their hydrocarbon chains are oppositely directed chain free fatty acids incorporated into a stratum corneum lipid model membrane based ceramide [AP] need to protrude into the adjacent layer

Table 5 Structure factors (SF) of the main phase of the SC lipid membrane containing either deuterated or protonated cerotic acid (CER[AP]/CHOL/d12CA or CA/ChS, 55/25/15/5) Table 6 at 8% D2O and 57% RH. Lamellar repeat distance d and the position x of the deuterated label of the FFA inside one lipid leaﬂet for each deuterated FFA containing membrane at 20 °C, 8% D O and 57% − 1 − 1 2 Diffraction order hQ,Å SF for CER[AP]/ Q,Å SF for CER[AP]/ RH. CHOL/CA/ChS CHOL/d12CA/ChS Phase of membrane d [Å] Position x of label, Å 1 0.143 −25.64±0.05 0.142 −20.22±0.04

2 0.285 +6.67±0.08 0.281 +6.78±0.08 d22BA: phase1_small 47.4±0.8 CD3-groups±0.01±0.04

3 0.432 −6.47±0.10 0.427 −7.78±0.10 d22BA: phase2_main 44.1±0.2 CD3-groups±4.08±0.02

4 0.577 +3.96±0.12 0.571 +3.49±0.13 d12CA: FFA-rich 56.5±0.6 CD2-groups±14.05±0.08

5 0.720 −4.93±0.17 0.711 −5.74±0.15 d12CA: main phase 43.8±0.4 CD2-groups±5.4±0.04 2202 A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203

Fig. 9. Armature reinforcement model of the SC lipid bilayer. Transformation of SC membrane from partly dehydrated state to fully hydrated by water excess. in order to arrange in the membrane size dictated by ceramide [AP]. [16] M.A. Kiselev, N.Y. Ryabova, A.M. Balagurov, S. Dante, T. Hauss, J. Zbytovska, S. Furthermore, the use of the deuterated free fatty acids confirmed the Wartewig, R.H. Neubert, New insights into the structure and hydration of a stratum corneum lipid model membrane by neutron diffraction, Eur. Biophys. J. 34 coexistence of free fatty acid rich phase and main phase. The formation (2005) 1030–1040. of this free fatty acids rich phase is a result of the decreased solubility of [17] M.W. de Jager, G.S. Gooris, I.P. Dolbnya, M. Ponec, J.A. Bouwstra, Modelling the the longer-chained free fatty acids within the membrane created by stratum corneum lipid organisation with synthetic lipid mixtures: the importance of synthetic ceramide composition, Biochim. Biophys. Acta 1664 (2004) 132–140. ceramide [AP]. The main forces which constrain the free fatty acids to [18] M.W. de Jager, G.S. Gooris, I.P. Dolbnya, W. Bras, M. Ponec, J.A. Bouwstra, Novel either interdigitate in the membrane or to separate as the free fatty acid lipid mixtures based on synthetic ceramides reproduce the unique stratum rich phase are caused by the very polar short-chain phytosphingosine- corneum lipid organization, J. Lipid Res. 45 (2004) 923–932. [19] M.W. de Jager, G.S. Gooris, M. Ponec, J.A. Bouwstra, Lipid mixtures prepared with type ceramide [AP] molecules, which therefore determine the stability well-defined synthetic ceramides closely mimic the unique stratum corneum lipid of this stratum corneum lipid model membrane. The presented results phase behavior, J. Lipid Res. 46 (2005) 2649–2656. support the phenomena of the armature reinforcement model of stratum [20] J.A. Bouwstra, G.S. Gooris, K. Cheng, A. Weerheim, W. Bras, M. Ponec, Phase behavior of isolated skin lipids, J. Lipid Res. 37 (1996) 999–1011. corneum lipid model membranes based on ceramide [AP] [55] and show [21] J.A. Bouwstra, G.S. Gooris, F.E. Dubbelaar, A.M. Weerheim, A.P. Ijzerman, M. Ponec, the importance of this short-chain ceramide for the nanostructure of Role of ceramide 1 in the molecular organization of the stratum corneum lipids, J. such model membranes. Additionally, the results of this study give new Lipid Res. 39 (1998) 186–196. material for the debate on the importance of the short-periodicity phase [22] J.A. Bouwstra, F.E. Dubbelaar, G.S. Gooris, A.M. Weerheim, M. Ponec, The role of ceramide composition in the lipid organisation of the skin barrier, Biochim. for the barrier function of the stratum corneum lipid bilayer. Biophys. Acta 1419 (1999) 127–136. [23] J.A. Bouwstra, G.S. Gooris, F.E. Dubbelaar, M. Ponec, Phase behavior of lipid mixtures based on human ceramides: coexistence of crystalline and liquid phases, Acknowledgements J. Lipid Res. 42 (2001) 1759–1770. [24] S.H. White, D. Mirejovsky, G.I. King, Structure of lamellar lipid domains and fi corneocyte envelopes of murine stratum corneum. An x-ray diffraction study, The work was nancially supported by the German Research Biochemistry 27 (1988) 3725–3732. Foundation (DFG) (NE 427/17-1). Further, financial assistance from [25] T.J. McIntosh, M.E. Stewart, D.T. Downing, X-ray diffraction analysis of isolated skin the Helmholtz Centre Berlin for Materials and Energy (Berlin, Germany) lipids: reconstitution of intercellular lipid domains, Biochemistry 35 (1996) 3649–3653. is also gratefully acknowledged. The authors would like to thank Evonik [26] T.J. McIntosh, Organization of skin stratum corneum extracellular lamellae: Goldschmidt GmbH (Essen, Germany) for the donation of CER[AP]. diffraction evidence for asymmetric distribution of cholesterol, Biophys. J. 85 (2003) 1675–1681. [27] B. Ongpipattanakul, M.L. Francoeur, R.O. Potts, Polymorphism in stratum corneum References lipids, Biochimica et Biophysica Acta (BBA), Biomembranes 1190 (1994) 122. [28] H. Farwanah, R. Neubert, S. Zellmer, K. Raith, Improved procedure for the separation of [1] P.W. Wertz, B. van den Bergh, The physical, chemical and functional properties of major stratum corneum lipids by means of automated multiple development thin-layer lipids in the skin and other biological barriers, Chem. Phys. Lipids 91 (1998) 85–96. chromatography, J. Chromatogr., B Anal. Technol. Biomed. Life Sci. 780 (2002) 443–450. [2] P.M. Elias, Epidermal lipids, barrier function, and desquamation, J. Invest. [29] A. Ruettinger, M.A. Kiselev, T. Hauss, S. Dante, A.M. Balagurov, R.H.H. Neubert, Dermatol. 80 (1983) 44s–49s (Suppl). Fatty acid interdigitation in stratum corneum model membranes: a neutron [3] P.M. Elias, D.S. Friend, The permeability barrier in mammalian epidermis, J. Cell diffraction study, Eur. Biophys. J. 37 (2008) 759–771. Biol. 65 (1975) 180–191. [30] J.A. Bouwstra, K. Cheng, G.S. Gooris, A. Weerheim, M. Ponec, The role of ceramides 1 [4] H.E. Bodde, I. Vandenbrink, H.K. Koerten, F.H.N. Dehaan, Visualization of in vitro and 2 in the stratum corneum lipid organisation, Biochim. Biophys. Acta 1300 (1996) percutaneous penetration of mercuric-chloride—transport through intercellular 177–186. space versus cellular uptake through desmosomes, J. Control. Release 15 (1991) [31] J.A. Bouwstra, J. Thewalt, G.S. Gooris, N. Kitson, A model membrane approach to 227–236. the epidermal permeability barrier: an x-ray diffraction study, Biochemistry 36 [5] P. Talreja, N.K. Kleene, W.L. Pickens, T.F. Wang, G.B. Kasting, Visualization of the (1997) 7717–7725. lipid barrier and measurement of lipid pathlength in human stratum corneum, [32] J.A. Bouwstra, G.S. Gooris, F.E. Dubbelaar, A.M. Weerheim, M. Ponec, pH, AAPS PharmSci 3 (2001) E13. cholesterol sulfate, and fatty acids affect the stratum corneum lipid organization, [6] G. Gray, H. Yardley, Different populations of pig epidermal cells: isolation and lipid J. Investig. Dermatol. Symp. Proc. 3 (1998) 69–74. composition, J. Lipid Res. 16 (1975) 441–447. [33] S.E. Friberg, D.W. Osborne, Interaction of a model epidermal lipid with a vegetable [7] G. Gray, H. Yardley, Lipid compositions of cells isolated from pig, human, and rat oil adduct, J. Dispers. Sci. Technol. 8 (1987) 249–258. epidermis, J. Lipid Res. 16 (1975) 434–440. [34] D. Kuempel, D.C. Swartzendruber, C.A. Squier, P.W. Wertz, In vitro reconstitution [8] G.M. Gray, R.J. White, Glycosphingolipids and ceramides in human and pig of stratum corneum lipid lamellae, Biochim. Biophys. Acta 1372 (1998) 135–140. epidermis, J. Investig. Dermatol. 70 (1978) 341. [35] G.C. Charalambopoulou, P. Karamertzanis, E.S. Kikkinides, A.K. Stubos, N.K. [9] H.J. Yardley, R. Summerly, Lipid composition and metabolism in normal and Kanellopoulos, A.T. Papaioannou, A study on structural and diffusion properties diseased epidermis, Pharmacol. Ther. 13 (1981) 383. of porcine stratum corneum based on very small angle neutron scattering data, [10] K. Raith, H. Farwanah, S. Wartewig, R.H.H. Neubert, Progress in the analysis of Pharm. Res. 17 (2000) 1085–1091. stratum corneum ceramides, Eur. J. Lipid Sci. Technol. 106 (2004) 561–571. [36] D. Kessner, A. Ruettinger, M.A. Kiselev, S. Wartewig, R.H. Neubert, Properties of [11] M. Ponec, A. Weerheim, P. Lankhorst, P. Wertz, New acylceramide in native and ceramides and their impact on the stratum corneum structure: part 2: stratum reconstructed epidermis, J. Invest. Dermatol. 120 (2003) 581–588. corneum lipid model systems, Skin Pharmacol. Physiol. 21 (2008) 58–74. [12] W.M. Holleran, M.Q. Man, W.N. Gao, G.K. Menon, P.M. Elias, K.R. Feingold, [37] D.L. Worcester, Neutron diffraction studies of biological membranes and Sphingolipids are required for mammalian epidermal barrier function. Inhibition membrane components, Brookhaven Symp. Biol., 1976, pp. III37–III57. of sphingolipid synthesis delays barrier recovery after acute perturbation, J. Clin. [38] N.P. Franks, Structural analysis of hydrated egg lecithin and cholesterol bilayers. I. Invest. 88 (1991) 1338–1345. X-ray diffraction, J. Mol. Biol. 100 (1976) 345–358. [13] L. Coderch, O. Lopez, A. de la Maza, J.L. Parra, Ceramides and skin function, Am. J. [39] M. Seul, M.J. Sammon, Preparation of surfactant multilayer films on solid substrates Clin. Dermatol. 4 (2003) 107–129. by deposition from organic solution, Thin Solid Films 185 (1990) 287–305. [14] M.E. Stewart, D.T. Downing, The omega-hydroxyceramides of pig epidermis are [40] D.L. Worcester, N.P. Franks, Structural analysis of hydrated egg lecithin and attached to corneocytes solely through omega-hydroxyl groups, J. Lipid Res. 42 cholesterol bilayers. II. Neutron diffraction, J. Mol. Biol. 100 (1976) 359–378. (2001) 1105–1110. [41] M.C. Wiener, S.H. White, Fluid bilayer structure determination by the combined [15] K.J. Robson, M.E. Stewart, S. Michelsen, N.D. Lazo, D.T. Downing, 6-Hydroxy-4- use of x-ray and neutron-diffraction. 1. Fluid bilayer models and the limits of sphingenine in human epidermal ceramides, J. Lipid Res. 35 (1994) 2060–2068. resolution, Biophys. J. 59 (1991) 162–173. A. Schroeter et al. / Biochimica et Biophysica Acta 1788 (2009) 2194–2203 2203

[42] J. Katsaras, D.S. Yang, R.M. Epand, Fatty-acid chain tilt angles and directions in [53] M.A. Fox, J.K. Whitesell, Organische chemie—grundlagen, mechanismen, dipalmitoyl phosphatidylcholine bilayers, Biophys. J. 63 (1992) 1170–1175. bioorganische anwendungen, Spektrum Akademischer Verlag, Heidelberg, [43] T. Hauss, S. Dante, N.A. Dencher, T.H. Haines, Squalane is in the midplane of the 1995. lipid bilayer: implications for its function as a proton permeability barrier, [54] D. Kessner, M. Kiselev, S. Dante, T. Hauss, P. Lersch, S. Wartewig, R.H.H. Neubert, Biochim. Biophys. Acta 1556 (2002) 149–154. Arrangement of ceramide [EOS] in a stratum corneum lipid model matrix: new [44] N.P. Franks, W.R. Lieb, The structure of lipid bilayers and the effects of general aspects revealed by neutron diffraction studies, Eur. Biophys. J. Biophys. Lett. 37 anaesthetics. An x-ray and neutron diffraction study, J. Mol. Biol. 133 (1979) 469–500. (2008) 989–999. [45] J.F. Nagle, S. Tristram-Nagle, Structure of lipid bilayers, Biochim. Biophys. Acta [55] M. Kiselev, Conformation of ceramide 6 molecules and chain-ﬂip transitions in the 1469 (2000) 159–195. lipid matrix of the outermost layer of mammalian skin, the stratum corneum, [46] G. Buldt, H.U. Gally, A. Seelig, J. Seelig, G. Zaccai, Neutron diffraction studies on Crystallogr. Rep. 52 (2007) 528. selectively deuterated phospholipid bilayers, Nature 271 (1978) 184. [56] L. Norlen, Skin barrier structure and function: the single gel phase model, J. Invest. [47] T. Hauss, G. Buldt, M.P. Heyn, N.A. Dencher, Light-induced isomerization causes an Dermatol. 117 (2001) 830–836. increase in the chromophore tilt in the M intermediate of bacteriorhodopsin: a [57] J.A. Bouwstra, G.S. Gooris, J.A. van der Spek, W. Bras, Structural investigations of neutron diffraction study, Proc. Natl. Acad. Sci. USA 91 (1994) 11854–11858. human stratum corneum by small-angle x-ray scattering, J. Invest. Dermatol. 97 [48] S. Dante, T. Hauss, N.A. Dencher, Insertion of externally administered amyloid beta (1991) 1005–1012. peptide 25–35 and perturbation of lipid bilayers, Biochemistry 42 (2003) 13667–13672. [58] G.C. Charalambopoulou, T.A. Steriotis, T. Hauss, K.L. Stefanopoulos, A.K. Stubos, A [49] H.S. Shieh, L.G. Hoard, C.E. Nordman, Crystal structure of anhydrous cholesterol, neutron-diffraction study of the effect of hydration on stratum corneum structure, Nature 267 (1977) 287–289. Appl. Phys., A Mater. Sci. Process. 74 (2002) S1245–S1247. [50] J. Huang, J.T. Buboltz, G.W. Feigenson, Maximum solubility of cholesterol in [59] S. Raudenkolb, S. Wartewig, R.H. Neubert, Polymorphism of ceramide 6: a phosphatidylcholine and phosphatidylethanolamine bilayers, Biochim. Biophys. vibrational spectroscopic and x-ray powder diffraction investigation of the Acta 1417 (1999) 89–100. diastereomers of N-(alpha-hydroxyoctadecanoyl)-phytosphingosine, Chem. [51] M.R. Ali, K.H. Cheng, J. Huang, Ceramide drives cholesterol out of the ordered lipid Phys. Lipids 133 (2005) 89–102. bilayer phase into the crystal phase in 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocho- [60] M.A. Kiselev, A.M. Kiselev, S. Borbely, P. Lesieur, Study of ethanol penetration through line/cholesterol/ceramide ternary mixtures, Biochemistry 45 (2006) 12629–12638. a model biological membrane by small-angle neutron scattering, Poverkhnost 6 [52] M.R. Brzustowicz, V. Cherezov, M. Zerouga, M. Caffrey, W. Stillwell, S.R. Wassall, (2006) 67–73. Controlling membrane cholesterol content. A role for polyunsaturated (docosa- [61] S. Motta, M. Monti, S. Sesana, R. Caputo, S. Carelli, R. Ghidoni, Ceramide hexaenoate) phospholipids, Biochemistry 41 (2002) 12509–12519. composition of the psoriatic scale, Biochim. Biophys. Acta 1182 (1993) 147–151.