Probing the Role of the Ceramide Acyl Chain Length and Sphingosine

Article pubs.acs.org/Langmuir

Probing the Role of the Ceramide Acyl Chain Length and Sphingosine Unsaturation in Model Skin Barrier Lipid Mixtures by 2H Solid-State NMR Spectroscopy Sören Stahlberg,† Barbora Školová,†,‡ Perunthiruthy K. Madhu,§,∥ Alexander Vogel,† Kateřina Vávrová,‡ and Daniel Huster*,†,§ †

Institute of Medical Physics and Biophysics, University of Leipzig, Härtelstr. 16-18, 04107 Leipzig, Germany Faculty of Pharmacy, Charles University in Prague, Heyrovského 1203, 50005 Hradec Králové, Czech Republic § Department of Chemical Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India ∥ TIFR Centre for Interdisciplinary Sciences, 21 Brundavan Colony, Narsingi, Hyderabad 500 075, India ‡

ABSTRACT: We investigated equimolar mixtures of ceramides with lignoceric acid and cholesterol as models for the human stratum corneum by differential scanning calorimetry and 2H solid-state NMR spectroscopy. Our reference system consisted of lignoceroyl sphingosine (Cer[NS24]), which represents one of the ceramides in the human stratum corneum. Furthermore, the effect of ceramide acyl chain truncation to 16 carbons as in Cer[NS16] and the loss of the C4 trans double bond as in dihydroceramide Cer[NDS24] were studied. Fully relaxed 2H NMR spectra were acquired for each deuterated component of each mixture separately, allowing the quantitative determination of the individual lipid phases. At skin temperature, the reference system containing Cer[NS24] is characterized by large portions of each component of the mixture in a crystalline phase, which largely restricts the permeability of the skin lipid barrier. The loss of the C4 trans double bond in Cer[NDS24] leads to the replacement of more than 25% of the crystalline phase by an isotropic phase of the dihydroceramide that shows the importance of dihydroceramide desaturation in the formation of the skin lipid barrier. The truncated Cer[NS16] is mostly found in the gel phase at skin temperature, which may explain its negative effect on the transepidermal water loss in atopic dermatitis patients. These significant alterations in the phase behavior of all lipids are further reflected at elevated temperatures. The molecular insights of our study may help us to understand the importance of the structural parameters of ceramides in healthy and compromised skin barriers.

■

INTRODUCTION Ceramides (Cer) belong to a family of sphingolipids that are constituents of the epidermal permeability barrier.1 In the human epidermis, Cer together with cholesterol (Chol) and free fatty acids form multilamellar membranes that fill the extracellular space of the uppermost epidermal layer, the stratum corneum (SC), and prevent the desiccation and entry of unwanted substances into the body.2,3 Alterations in Cer content and proportions contribute to barrier defects accompanying skin diseases such as atopic eczema, psoriasis, and ichthyoses.2,3 For the therapy of these conditions, topical supplementation of Cer or their less-expensive analogs is an interesting opportunity,4 which is, however, limited by the fact that the structural requirements and behavior of Cer are not fully known.5 DihydroCer, i.e., the sphingolipid species that lack the C4 trans double bond such as nonhydroxy acyl dihydroCer (Cer[NDS]), are biosynthetic precursors of Cer (Figure 1).6 They have been detected in the SC only recently,7 and their function in skin barrier homeostasis is unknown. The role of dihydroCer in cellular and extracellular processes is particularly interesting because since the 1990s the C4 trans double bond has been presumed to be essential to a variety of biological functions of Cer, and the corresponding dihydroCer were © 2015 American Chemical Society

described to lack activity. Later studies showed that the longchain dihydroCer are not inactive, although their effects differ from those of Cer.6 Some studies suggest that a specific Cer/ dihydroCer ratio or sphingosine/dihydrosphingosine ratio might be crucial for proper skin barrier function.8−10 However, a direct comparison suggested that the presence of the trans double bond in Cer is not essential to their barrier function: the permeability of model SC lipid membranes containing nonhydroxy acyl species of dihydroCer (Cer[NDS24]11) or Cer (Cer[NS24]) was either comparable or only slightly greater in the membranes containing Cer[NDS24].12 The barrier properties of sphingolipids are also markedly influenced by their acyl chain length. According to the Cer synthase involved in their formation, the Cer acyls vary from long (13−19 carbons) to very long (20−26 carbons) to ultralong (>28 carbons).13 In atopic dermatitis patients, increased levels of long Cer species (palmitoyl sphingosine, Cer[NS16]) at the expense of the very long acyl Cer have been found, and this change has been correlated with the skin barrier properties.14,15 This Cer chain-length shift is most likely caused Received: February 27, 2015 Revised: April 13, 2015 Published: April 14, 2015 4906

DOI: 10.1021/acs.langmuir.5b00751 Langmuir 2015, 31, 4906−4915

Article

Langmuir

Cer[NS24], and Cer[NDS24]in an equimolar mixture with LA and Chol.

■

MATERIALS AND METHODS

Materials. Chol, sphinganine (dihydrosphingosine), sphingosine, Cer[NDS24], and Cer[NS24] were purchased from Avanti Polar Lipids (Alabaster, AL). Chol(2,2,3,4,4,6)-d6 was purchased from Cambridge Isotope Laboratories, Inc. (Tewksbury, MA), and palmitic acid-d31 and lignoceric acid-d47 (LA-d47) were purchased from C/D/N Isotopes, Inc. (Pointe-Claire, Canada). Cer[NS16] and all other chemicals were purchased from Sigma-Aldrich (Darmstadt, Germany) and used without further purification. Synthesis of Cer with Perdeuterated Acyls. N-((2S,3R)-1,3Dihydroxyoctadecan-2-yl)tetracosanamide-d47 [Cer[NDS24]-d47): LAd47 (0.14 mmol), 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (0.14 mmol), and N-hydroxysuccinimide (0.14 mmol)] was dissolved in dry CH2Cl2 under nitrogen, and then dihydrosphingosine (0.14 mmol) in CH2Cl2 was added at 0 °C. The reaction solution was allowed to warm up to room temperature and was stirred overnight. Next, the reaction mixture was washed with 0.1 M HCl and water. The organic phase was dried over Na2SO4, concentrated under reduced pressure, and purified by silica gel column chromatography using CHCl3/methanol 50:1 as the mobile phase. Yield: 30%, white crystals, mp = 98−101 °C, Rf = 0.43 (CHCl3/methanol 10:1). IR (ATR): ν 3397, 2917, 2851, 2193, 2089, 1623, 1471, 1091, 721 cm−1. 1H NMR (CDCl3, 500 MHz): δ 6.54 (br s, 1H), 3.99 (dd, J = 11.4, 3.6 Hz, 1H), 3.86−3.80 (m, 1H), 3.79−3.72 (m, 2H), 2.92 (br s, 2H), 1.54−1.50 (m, 2H), 1.35−1.20 (m, 26H), 0.88 (t, J = 6.7 Hz, 3H). 13C NMR (CDCl3, 125 MHz): δ 173.8, 74.1, 62.4, 53.9, 38.6, 34.5, 31.9, 29.7, 29.7, 29.6, 29.6, 29.4, 26.0, 22.7, 19.2, 14.1, 13.7. MS (ESI+) m/z 700.1 (M + H+), 682.1 (M + H+ − H2O). Cer[NS24]-d47 and Cer[NS16]d31 were prepared by a condensation of sphingosine with LA-d47 or palmitic acid-d31.17 Thin-layer chromatography was performed on Merck (Darmstadt, Germany) aluminum plates with silica gel 60 F254. Melting points were recorded on a Büchi B-545 apparatus (Büchi Labortechnik AG, Flawil, Switzerland) and are uncorrected. IR spectra were recorded on a Nicolet 6700 (Thermo Scientific, Waltham, MA) equipped with a single-reflection MIRacle ATR ZnSe crystal (PIKE Technologies, Madison, WI). NMR spectra were recorded on a Varian VNMR S500 spectrometer (Palo Alto, CA). Chemical shifts were indirectly referenced to tetramethylsilane via the solvent signal. Electrospray ionization mass spectrometry was conducted using an Acquity UPLC with MS/MS Quattro Micro detection (Micromass+Waters, Milford, MA). Sample Preparation. For each ceramide, three samples were prepared: Cer (deuterated)/LA/Chol (1:1:1 mol/mol/mol); Cer/LAd47 (deuterated)/Chol (1:1:1); and Cer/LA/Chol-d6 (deuterated) (1:1:1). Aliquots of Cer, fatty acid, and Chol were dissolved in chloroform/methanol (2:1), and the solvent was evaporated using a rotary evaporator. Afterward, the sample was dissolved in cyclohexane and lyophilized overnight. The obtained powder was hydrated with 50 wt % deuterium-depleted water. For homogenization, samples were first placed in 4 mm NMR rotors, which were used as sample containers and sealed for static 2H NMR measurements to avoid dehydration. Next, each sample was frozen in liquid nitrogen and heated to 80 °C 10 times. Because a recent study showed that 4−10 h is required for the formation of orthorhombic domains of fatty acid in samples with Cer[NS18],28 all samples were incubated for 24−30 h at 22 °C so that the observed differences in the NMR spectra were not caused by the different kinetics in phase formation. Although other studies suggested that the formation of crystalline phases in Cer/palmitic acid/Chol mixtures containing either bovine brain Cer type III (acyl chain lengths, mainly C18, C24, and C24:1) or Cer[NS16] required longer periods,26,29 our previous study on CerNS-based systems showed that the IR spectra did not change with incubation longer than 24 h. In addition, no changes in permeability to two model drugs of such model lipid membranes were found between 1 and 14 days.20 NMR spectra of

Figure 1. Structures and biosynthesis of the investigated Cer and dihydroCer. DihydroCer are generated by a de novo biosynthetic pathway from L-serine followed by an introduction of the C4 trans double bond by desaturase to form Cer (panel A). The structures and abbreviations11 of the investigated lipids are given in panel B.

by a down regulation of elongases as described in a murine atopic dermatitis model.16 SC membrane models suggested that Cer[NS24] but not Cer[NS16] prefers the extended conformation in the skin lipid membranes in which a fatty acid is associated with the Cer[NS24] acyl.17 In addition, shortchain C2−C12 Cer increase the permeability of the skin18,19 and model lipid membranes,20 and they also penetrate to different depths of the epidermis after their topical application to the human skin compared to long-chain Cer.21 On the other hand, phytosphingosine-based Cer with α-hydroxylated C18 acyl was found to increase the barrier function of the SC model membranes and decrease the permeation of drugs.22 Here, we investigate the role of two (patho)physiologically relevant structural parameters in Cer, namely, the C4 trans double bond in the sphingoid chain and acyl chain length (C24 or C16), in the thermotropic phase behavior of equimolar mixtures of Cer, lignoceric acid (LA, tetracosanoic acid, C24:0), and Chol by differential scanning calorimetry (DSC) and 2H solid-state nuclear magnetic resonance (NMR) spectroscopy. The structures of the Cer studied are also given in Figure 1. 2 H solid-state NMR represents a powerful method for studying the structure, dynamics, and phase state of lipid assemblies. Cer-containing mixtures that typically contain free fatty acids, Chol, and phospholipids have been studied by the method in great detail.23−27 However, deuterated Cer molecules have become available only recently, and only a few studies have been conducted that analyzed the 2H NMR spectra of the individual components of SC model systems.25,26 In the study by Brief et al., the thermotropic phase behavior of an equimolar mixture of Cer[NS16], palmitic acid, and Chol was investigated.26 By switching the 2H label between the palmitic acid of the Cer and the free palmitic acid, the molecular structure and dynamics as well as the phase behavior of these components of the system could be described individually. Here, we follow this approach and investigate three biologically highly relevant Cer moleculesCer[NS16], 4907

DOI: 10.1021/acs.langmuir.5b00751 Langmuir 2015, 31, 4906−4915

Article

Langmuir identical samples remained very similar over time periods of several weeks. In the NMR spectrometer, spectra were recorded at five temperatures in the following order: 25, 32, 50, 65, and 80 °C. Samples for DSC were prepared as suspensions in 10 mM pyridine buffer at pH 5.5 (154 mM NaCl), heated to 95 °C in an ultrasound bath, and then slowly (over ∼4 h) cooled to room temperature. This heating−cooling cycle was repeated four times. Differential Scanning Calorimetry. DSC measurements were performed with a nanodifferential scanning calorimeter (TA Instruments, Lindon, UT). Samples were degassed prior to injection and equilibrated for 10 min before each heating and cooling scan and then scanned from 20−120 °C at a scan rate of 1 K/min relative to pyridine buffer, equilibrated, and cooled at the same rate. Three heating scans were collected for each sample. Because these scans were identical, we report only the results of the first scan. DSC curves were analyzed using NanoAnalyze 2.0 (TA Instruments, Lindon, UT). 2 H NMR Spectroscopy. NMR measurements were performed on a Bruker AVANCE 750 WB NMR spectrometer (Bruker BioSpin, Rheinstetten, Germany) at a resonance frequency of 115.1 MHz using a probe with a 5 mm solenoid coil at a spectral width of ±250 kHz using quadrature phase detection, a phase-cycled quadrupolar echo sequence30 with two 2−2.5 μs 90° pulses separated by a 30 μs delay. The recycle delay was 50 s for all samples. Samples were measured at temperatures of 25, 32, 50, 65, and 80 °C. Spectra were processed using a program written in Mathcad (MathSoft, Cambridge, MA).31 Typically, one NMR spectrum was accumulated per sample and selected samples were repeated to confirm reproducibility. 2 H NMR Line Shape Simulations. 2H NMR line shapes were simulated by assuming a superposition of N 2H NMR Pake doublets, scaled by the appropriate order parameter (SCD) and typically one isotropic line. The rigid limit quadrupolar coupling was 125.25 kHz, assuming a quadrupolar coupling constant of χ = 167 kHz for the C−2H bond. Time domain data were simulated as free induction decays (FID) with an increment of the powder angle θ of 0.0625° according to N

FID(t ) =

⎛⎛

∑ ⎜⎜∫ i=1

⎝⎝

90 °

θ=0

⎞ exp(tLBi π )⎟ ⎠

Figure 2. DSC heating curves of the hydrated Cer/LA/Chol mixtures in buffer (pH 5.5) collected at a 1 K/min scan rate. The numbers indicate the phase-transition temperatures.

Cer[NDS24]/LA/Chol mixture exhibits two resolved peaks at temperatures of 32 and 69 °C. It may seem counterintuitive to observe a lower transition temperature for the very long chain Cer[NS24]-containing mixture than for the mixture containing its shorter homologue Cer[NS16]. However, the cohesive forces in the lipid mixtures are also influenced by the Cer molecular shape, packing, and mixing with Chol and/or free fatty acid. Thus, the energy required to overcome the cohesive forces in membranes may actually be lower for longer-chain Cer, although the phase transitions of pure long Cer occur at similar temperatures.35 These phase-transition temperatures are consistent with those obtained by infrared spectroscopy of Cer/ LA/Chol mixtures with 5% Chol sulfate, which were 69 °C for Cer[NS16],17 59 °C for Cer[NS24],18,20 and 67 °C for Cer[NDS24].12 2 H NMR of the Cer[NS16]/LA/Chol Mixture. The 2H NMR spectra of Cer[NS16]/LA/Chol are shown in Figure 3. Each column depicts the 2H NMR spectra of one deuterated species of the mixture, i.e., in the left column, the Cer[NS16]d31; in the middle column, the LA-d47; and in the right column, the Chol-d6 is labeled. The rows represent the temperature starting at 25 °C on the bottom to 80 °C in the top row. Thus, the figure summarizes the phase behavior of this specific mixture. At 25 °C, the 2H NMR spectrum of Cer[NS16] is very reminiscent of that of a gel phase as known from phospholipid membranes. Small differences in the NMR line shape of pure phospholipid membranes in the gel phase may be attributed to differences in rotational diffusion rates.26 This result suggests that the majority of Cer[NS16] forms a gel phase. The LA spectrum is dominated by the crystalline phase with some small proportion of the acid being more fluid. The NMR spectrum of the Chol in the mixture also displays the characteristic features of a crystalline phase. Numerical simulations of the 2H NMR spectra allow us to quantitatively describe the phase composition of these mixtures. Simulated 2H NMR spectra are also shown in Figure 3 (red lines). Because it has been reported that the crystalline portions of SC mixtures feature long relaxation times of up to 10 s,34 the 2 H NMR spectra had to be recorded at a long relaxation delay of 50 s. Therefore, we could determine the relative proportion of each phase in the mixture directly from the line-shape simulation of the 2H NMR spectra. The result of this analysis is given in Table 1. At 25 °C, the 2H NMR spectra of each individual component can be described by a superposition of two or three line shapes, each characteristic of one lipid phase. The majority (97%) of the Cer[NS16] forms a gel phase. The LA 2H NMR spectrum can be simulated by a crystalline (73%) and a more fluid

⎞ ⎛3 1 ⎞ dθ cos⎜ χ (3 cos2 θ − 1)SCDi t ⎟sin θ ⎟ ⎝4 2 ⎠ ⎠

(1)

In eq 1, LB is the line-broadening factor and θ is the angle for powder averaging. For spectra that showed some magnetic-field-induced orientation, the spherical distribution function sin θ was replaced by an ellipsoidal distribution function according to32 ((2πc2 sin θ)/(sin2 θ + (c2/a2)cos2θ)2), where a and c are the eccentricities. The gel-phase spectrum was simulated using a superposition of two C−2H bond vectors undergoing two-site exchange with a hop angle of 120° for each methylene group in the chain and the NMR spectrum of methyl groups undergoing fast rotation. The two-site exchange line shape was simulated using NMR-WEBLAB,33 and all other NMR spectra were simulated using programs written in MathCad. The relative proportions of the individual phases observed in the 2H NMR spectra were directly determined from these fits. Because 2H NMR spectra were acquired at a relaxation delay of 50 s to account for the long relaxation times (up to 10 s) of the crystalline phase,34 the relative contributions of each phase were determined from the relative proportion of each individual line shape for each phase that gave rise to the final NMR spectrum of the respective sample.

■

RESULTS DSC. To get a first impression of the phase behavior of the ternary lipid mixtures, DSC measurements on aqueous dispersions of the lipids were conducted. The DSC heating traces are shown in Figure 2. The equimolar mixture of Cer[NS16]/LA/Chol shows three relatively well resolved peaks at 52, 68, and 71 °C. The Cer[NS24]/LA/Chol mixture shows peaks at 30 and 62 °C and a smaller peak at 67 °C. The 4908

DOI: 10.1021/acs.langmuir.5b00751 Langmuir 2015, 31, 4906−4915

Article

Langmuir

Figure 3. 2H NMR spectra of the equimolar mixture of Cer[NS16]/LA/Chol at varying temperatures and 50 wt % hydration. The left column shows the 2H NMR spectra of the Cer[NS16]-d31, the middle column depicts the 2H NMR spectra of the LA-d47, and the right column displays the 2H NMR spectra of the Chol-d6 in the mixture. Spectra were acquired with a relaxation delay of 50 s. The red lines illustrate the best-fit numerical simulations.

Table 1. Quantitative Analysis of the Phase Composition of Equimolar Ternary Cer[NS16]/LA/Chol Mixtures Determined from Numerical Simulation of the 2H NMR Spectra Shown in Figure 3a Cer[NS16]

a

lignoceric acid

cholesterol

temperature (°C)

gel (%)

fluid (%)

isotr (%)

cryst (%)

fluid (%)

isotr (%)

cryst (%)

fluid (%)

isotr (%)

25 32 50 65 80

97 95 77 44 0

3 5 23 56 16