Methodological Considerations for Lipid and Polar Component Analyses in Human Skin Stratum Corneum

Abstract

Collection of skin very top layer, called stratum corneum, by tape stripping and the analysis of stratum corneum components by mass spectrometry provides multiple advantages for clinical studies that aim to understand the origins of allergic skin diseases and food allergy. However, such a methodology has multiple challenges on the way of complex stratum corneum analysis when molecules of different polarity are needed to be analyzed from minimal amount of skin tape strips. This review provides an overview of current knowledge about lipid and polar molecules in the skin, discusses challenging aspects of sample processing when dealing with skin tape strips, and provides some guidance towards approaches that generate complex, quantitative, normalized to total sample protein data that fit best the purpose of analysis of stratum corneum components for the purpose of clinical trials.

Introduction

Human skin has very important functions. First, it provides a physical barrier that prevents the loss of water from inside our body to the atmosphere. Second, it protects against penetration of allergens, bacteria, and viruses into our body. These barrier properties of the skin are made possible through a highly orchestrated process of terminal differentiation of keratinocytes, the major skin-specific cells, that undergo dramatic biochemical transformations to produce well organized, multilayered, highly hydrophobic lamellae structures that, together with residual dead keratinocytes (corneocytes) form an impermeable barrier on the surface of the skin. This very top layer of the skin is called stratum corneum, and for a long time it was not considered to bare significant biochemical information that can inform about immunological events happening deeper in the skin. However, it was noted that skin lipid and protein composition changes substantially in lesional skin areas in patients with skin diseases, such as atopic dermatitis (AD) and psoriasis [1–4], and further work identified strong association between the impaired expression of selected proteins, such as filaggrin, and the risk of developing AD, food allergy, and eczema herpeticum [5, 6]. While early work allowed an advanced progress in understanding the link between hyperactivated immune response and changes in the process of keratinocyte terminal differentiation that lead to abnormal skin development [5, 7, 8], the exact details of complex changes in associated gene, protein, and lipid expression in the skin became actively explored only recently with the development of modern sequencing and mass spectrometric tools. Within skin components, lipids are probably the last to gain advantage from modern analytical techniques, and the last ten-fifteen years brought tremendous progress in understanding the diversity and uniqueness of skin lipids. In this review, we will focus on methodological aspects of lipid analyses in stratum corneum and will discuss the applicability of stratum corneum component analysis as a window towards events underlying the failure of skin barrier properties in allergic skin diseases.

Lipid and polar molecules in stratum corneum

The very first comprehensive analysis of lipids in human stratum corneum was published in 1983 by Dr. Elias group [9]. This study described the unusual lipid composition in stratum corneum that contains very little polar lipids, the abundance of free cholesterol, free fatty acids, ceramides, and appreciable amount of cholesterol sulfate. Parallel work has revealed that epidermal lipids undergo gradual transformations during transition from stratum basale, where polar lipids comprise the majority of lipids, to stratum corneum, where polar lipids are virtually absent and free fatty acids, free cholesterol, and ceramides dominate [10]. This latter study also demonstrated the increase in the levels of glycosylceramides in stratum granulosum and their disappearance in stratum corneum that suggests their importance for a proper genesis of stratum corneum ceramides. It also was the first study to show the separation of otherwise single or double ceramide band on thin layer chromatography (TLC) plate into six different bands without further documentation of their identity [10]. Now we know that skin ceramides are very complex and currently are separated into twelve different subgroups based on the combination of the type of sphingoid base and N-linked fatty acids in ceramide molecule (Fig. 1). Due to that complexity, it is currently common to use abbreviations shown in Fig. 1 instead of systematic or Lipid Maps designated names for skin-specific ceramides.

Figure 1. Classification of skin ceramides.

Skin lipids are currently classified into twelve subgroups based on the combination of Long Chain Base (LCB) and the type of N-linked fatty acid (FA). Shown abbreviations are used to designate a specific group of skin ceramides instead of systematic or Lipid Maps designated names to define which group of ceramides is currently discussed. It is estimated that stratum corneum ceramides are comprised of at least 300 different molecular species; however, this number is currently challenged and can be many thousands.

Lipids in stratum corneum are mostly organized in well-defined lamellae structures called cornified envelope that is formed by differentiated keratinocytes at the very last stages of keratinocyte terminal differentiation and cell death [9, 10]. In healthy skin, such lamellae are uniformly organized and tightly packed, and provide a physical barrier for water loss as well as for penetration of allergens and bacteria. However, in atopic skin diseases, their integrity is compromised due to either genetic abnormalities (for example, null mutations in the filaggrin gene), or changes in the expression of proteins or enzymes that control lipid metabolism, or protein biotransformations as a result of hyperactivated immune response. [6, 11]. Out of twelve subgroups of skin ceramides, one type of skin ceramides, called esterified omega-hydroxyacyl-sphingosine, or EOS ceramides, together with EODS and EOP ceramides, fulfill a particularly important function. They serve as precursor molecules to generate a lipid scaffold over proteins in the lamellae structures through transglutaminase mediated cross-linking of protein glutamines to a ceramide terminal fatty acid hydroxy group after oxidation and removal of omega hydroxy-linked linoleic acid [12–14]. In addition, free EOS ceramides are theoretically able to partition between both bilayers of the lamellae due to extreme length of their hydrophobic tail and, thus, increase overall tightness of the proteo-lipid structure, while some mathematical modeling suggests the folding of O-linked linoleic acid in the intermembrane space [15].

Another important characteristic of skin ceramides, in addition to the number of different sphingoid bases, is the abundance of very long-chain fatty acids (C26-C34) in their structure [16, 17]. These very long-chain fatty acids provide additional rigidity and hydrophobicity to the lipid structure that is required to ensure proper barrier properties of the stratum corneum. In fact, one of the factors that lead to skin barrier impairment in AD, in addition to an overall decrease in lipid to protein ratio [18], is an overall shortening of ceramide-linked fatty acids [2, 19–22] due to the inhibition of expression of fatty acid elongases ELOVL3 and ELOVL6 by type-2 cytokines [21, 23]. Therefore, the task of proper identification and quantitation of stratum corneum lipid components carries special challenges due to the complexity of stratum corneum lipids and their variability that relates to the state of skin barrier impairment.

Within polar components of stratum corneum, so called Natural Moisturizing Factor (NMF) is probably the most important to characterize. NMF is a mixture of different amino acids and peptides generated by proteases and other enzymes during stratum corneum maturation and serves, in part, to maintain acidity and to retain moisture in the skin. Within components of NMF, pyroglutamic acid (PCA) and urocanic acid (UCA) are the most important as in the skin, they originate mostly from filaggrin protein that is critical for the assembly of proteo-lipid complexes and is rich in glutamine and histidine that serve as precursors for PCA and UCA, respectively [24, 25]. Therefore, their content in stratum corneum indirectly points at the expression level of filaggrin in the skin that can be affected by either mutations in Flg gene or by type 2 cytokines [24, 25].

Collection of stratum corneum and extraction of lipids

While collection of the very top layers of the skin seem to be a trivial task, it took a significant effort to develop materials and procedures that allow standardized collection of stratum corneum to ensure repetitive analyses over time and at different locations. The very first procedure to collect skin lipids was just to wash parts of human body with ethanol and collect it in a wide basin [26]. Brief washing of the skin with ethanol does not harm the host but elutes lipids only from very top stratum corneum layers. The thickness of human stratum corneum varies depending on the region of the body, with multiple layers of corneocytes in the volar area of the forearm that is frequently used for stratum corneum collection in clinical studies, until it transitions to stratum granulosum with less differentiated and still alive cells. The washing of body regions with solvents was quickly replaced with skin tape stripping (STS) that involves application of a polymeric tape with adhesive component that strips layers of corneocytes upon its removal. However, STS seeming simplicity holds several challenges [27] that should be considered. One of the major challenges is that each strip, when sequentially applied to the same region of the skin, removes different amounts of corneocytes, the amount of material removed depends on the STS layer and individual differences between subjects. In fact, individual differences between subjects are so big that the amount of removed stratum corneum even for the same strip number can be as high as 10 times between subjects. Therefore, maximum efforts should be dedicated to the standardization of STS procedure. The use of a pressure device that provides uniform application (for example, D-Squame pressure instrument, Clinical and Derm, Dallas, TX) is strongly recommended. The second major challenge is the interference of adhesive components on the skin tape with subsequent analyses, as all organic solvents of different polarities (from alcohols to hexane) either dissolve adhesives or extract their components. About twenty years ago, Clinical and Derm (previously CuDerm, Dallas, TX) introduced D-Squame sampling disks for stratum corneum sampling. Yet, their adhesive is prone to the same damaging effects from organic solvents. While several procedures have been developed for the analysis of either polar [28] or lipid [2, 16, 17, 19–21, 29–31] components of the skin, they all cannot be used for the analysis of both polar and lipid components from the same sample due to the interference of adhesive material with final analyses. Importantly, these processing protocols often satisfy only relative abundance type of analyses but not the quantitative analyses, as most of these protocols lack normalization to the amount of sample collected. Even when data are presented per mg protein or μg of stratum corneum, many published protocols lack important details needed to conclude about sample recovery and, most importantly, how exactly protein content was measured. To overcome that obstacle, our group has created a procedure that allows the analysis of both lipid and polar components from two or even one STS with all data normalization per total protein amount in the sample [21,32,33]. It also allows quantitation of protein bound lipid components in the same sample. Together with a standardized sample collection, such a processing is ideal for clinical trials as it requires minimum amount of STS (1–2), provides quantitative data for all groups of molecules of interest, and allows comparison of samples distant in time and location.

The critical difference of our STS processing is that we exclude adhesive components at the very first step as stratum corneum is removed from STS by scraping (for example, in a 6-well cell culture plate) in a solution of water : methanol (9:1, v/v, 2.ml) that becomes part of further extraction solution. The suspension of corneocytes is transferred into in a glass tube then well is washed twice with methanol (2×1.15 ml) to collect residual stratum corneum particles. Then, small amount of chloroform (0.25 ml), and appropriate internal standards are added. After lipid extraction (1 hour minimum), chloroform (2.ml) and 2% formic acid (0.45 ml) are added to induce phase separation. In fact, this procedure can be considered as a variant of Bligh and Dyer extraction protocol [34]. After vortexing and centrifugation, bottom chloroform phase is carefully collected and the residual solvents with protein interphase are cleaned by adding chloroform (2.5 ml), tube vortexing, centrifugation, and collection of the chloroform phase that is combined with the first portion. Second extraction is important to clean protein interphase from any residual lipids, so they do not interfere with the analysis of protein-bound ceramides. Next, the upper phase with polar components is carefully collected. The extraction tube containing residual proteins is subjected to basic hydrolysis, after evaporation of residual solvents under a stream of nitrogen, with 1N NaOH (0.5 ml) at 80°C for three hours. After neutralization with 1N HCl (0.5 ml), protein hydrolysate is analyzed for protein content then processed for another Bligh and Dyer extraction with the addition of second portion of internal standards to analyze protein-bound ceramides and fatty acids. This procedure, while lengthy and tedious, ensures quantitative and normalized to sample protein analysis of lipids, polar, and protein-bound components with minimal losses of material during processing steps with just one or two STS (Fig. 2).

Figure 2.

Skin tape strip processing protocol that allows simultaneous analysis of free lipids, polar components, and protein-bound lipids with data normalization per sample total protein content.

Mass spectrometric analysis of stratum corneum lipid and polar components

Mass spectrometric analysis of lipid and NMF components of stratum corneum is now well developed and actively used [16, 17, 24, 31, 35–45]. However, the complexity of stratum corneum ceramides puts certain restraints on some instrumentation and requires good chromatographic separation of molecules due to multiple isobaric overlaps. What is not well appreciated is that in contrast to ceramides in most tissues and cells, which have C18-sphingosine and C18-dihydrosphingosine as the major sphingoid bases in ceramide structures, stratum corneum ceramides are rich not only in C18-sphingoid bases (Fig. 1) but also in sphingoid bases with higher number of carbon atoms, including the presence of substantial amount of sphingoid bases with the odd number of carbons [31]. Kihara group has nicely uncovered the metabolic pathway that leads to the generation of odd-chain fatty aldehydes and corresponding odd-chain fatty acids from sphingolipid turnover. It involves phytoceramide metabolism to phytosphingosine-1-phosphate and its degradation by S1P lyase that results in the formation of pentadecanal instead of Δ2-hexadecenal or hexadecanal that are formed from sphingosine-1-phosphate and dihydrosphingosine-1-phosphate, respectively [46]. The pentadecanoic fatty acid can enter into sphingolipid biosynthesis and form C17-sphingoid bases. What is currently not clear is why serine palmitoyltransferase, enzyme that catalyzes the very first step of sphihgolipid de novo biosynthesis and is usually very restrictive to the chain length of involved Acyl-CoAs with the preference towards palmitic acid, becomes promiscuous in differentiating keratinocytes and accepts Acyl-CoAs with a chain length up to 26 carbon atoms. Furthermore, even being already published in several structural studies [31, 42, 43], the presence of ceramides with extended long chain sphingoid bases is not actively discussed in the clinical literature. Meanwhile, ceramides with long chain sphingoid bases should be beneficial for stratum corneum barrier properties from biophysical point of view, as they will form more rigid structures in comparison with C18-based ceramides. Furthermore, ceramides with C20- and even more with C22-based sphingosine are primarily linked with very long-chain fatty acids (C26-C32), that in overall makes those ceramides highly hydrophobic and rigid (see below and Fig. 6).

Figure 6. LC-MS/MS detection of NMF components in human stratum corneum.

Two STS were processed for detection of NMF components and amino acids. LC separation of cis/transUCA, PCA, and other amino acids was achieved using an Acquity UPLC BEH Amide (2.1 × 100 mm, 1.7 μm particle size) column using a gradient from acetonitrile (Solvent A) to methanol:water:formic acid (65:35:0.5, with 5 mM ammonium formate) (Solvent B). All amino acids were detected in positive ions mode using AB Sciex 6500QTRAP instrument. Note the prevalence of the signal for cis-/trans-UCA due to their preferred ionization efficiency at employed chromatographic conditions. Dotted line represents the gradient between solvents A and B (% is shown).

Sadowski et al. [42] performed a comparison of relative composition of lipids collected by the D-Squame tape strips from healthy volunteers and demonstrated that lipid composition in stratum corneum varies substantially within the very first four strips, but then becomes stable with the exception for triglycerides. As for the ceramides, their relative percentage within lipids of stratum corneum becomes more or less stable after the third strip. This study is important as it provides insight into the experimental design for clinical trials and demonstrates the need to refrain from the use of STS 1–3 for lipid analyses to decrease inter-sample variability. While this study did not analyze NMF components, it is logical to assume that NMF concentration in superficial STS layers will be substantially reduced by washing due to their polar nature. Therefore, it would be safe also to avoid strips 1–3 for the analysis of NMF.

The works by Sandra [17] and Voegel [31] groups are probably the most comprehensive in the description of mass spectrometric analyses of stratum corneum ceramides and provide vast information regarding ceramide mass spectrometric identification. Both groups point out the superiority of ceramide analysis in positive ions mode in comparison to negative ions mode and supplement their reports with multiple mass spectra examples and additional information that facilitates setting up these assays. Both works provide information regarding the presence of ceramides with sphingoid bases of different chain length from C16 to C26, including sphingoid bases with the odd chain length. According to their work, the most abundant chain length within all sphingoid bases is C18-base, while in EOS-Ceramides, C20-sphingoid base is the predominant one in the stratum corneum. Also, the relative prevalence of sphingoid bases with different chain lengths depends on the species analyzed [31]. It is clear that more information regarding structural diversity of stratum corneum ceramides will become available in subsequent studies. The extension of the currently accepted classification of skin ceramides from 12 groups (Fig. 1) into 25 groups due to discovery of β-hydroxy fatty acids, ω-hydroxy fatty acids, and 4,14-sphingadiene as sphingoid base within free stratum corneum lipids is an example of evolution of our knowledge about structural diversity of ceramides in stratum corneum [41]. Recent discovery of 1-O-acylceramides [47] even further expands the complexity of ceramides in stratum corneum and enzymatic pathways involved in their generation.

The development of an assay for stratum corneum ceramides is a tedious task. While a survey scan for any lipid that has a free hydroxy group in the lipid extract from human stratum corneum (all ceramides have at least one free hydroxy group, except for 1-O-acyceramides) by performing a neutral loss of 18Da scan seems to be not too complex and provides a rough estimate of relative proportion between EOS-ceramides and most other ceramides (Fig. 3), the full complexity of ceramide family in stratum corneum becomes apparent when performing further structural analyses. For example, a simple product ions scan from the m/z 678.5 that usually corresponds to N(26:0)S(18) ceramide in positive ions, reveals that this molecular ion gives rise to at least three different fatty acids (24:0, 25:0, 26:0), sphingosines with 18, 19, 20 carbon atoms, phytosphingosines with 18 and 20 carbon atoms (detected at m/z 678.5 due to in-source loss of water during ionization), and also, potentially, to dihydrosphingosine with 19 carbon atoms and 6-hydroxysphingosine with 18 carbon atoms (Fig. 3A). The inverse experiment, when looking at precursor ions for the m/z 264.4 that corresponds to sphingosine with 18 carbon atoms without two water molecules, reveals that predominant NS ceramides are molecular species with 26:0, 24:0, 28:0, and 30:0 fatty acids, but there is also a substantial amount of ceramides with 24:0-OH, 26:0-OH, 28:0-OH, 30:0-OH, and 32:0-OH fatty acids. Furthermore, this experiment also reveals the presence of at least two EOS ceramide molecular species with C18-sphingosine as a backbone (Fig. 3B). Due to that combinatorial complexity and the progress in revealing new sphingoid bases and fatty acids within ceramide molecules, it is currently estimated that the total number of different ceramide molecular species in the skin can be not even hundreds but many thousands of molecules. Such a diversity of ceramides in the skin suggests the need for specialized methods of ceramide analyses depending on the objectives of each study, where structure elucidation studies require the most coverage of different molecular species, and clinical studies covering, probably, only the most abundant ceramide species that characterize the outcome of clinical interventions. Next, improvements in data visualization may provide additional information that might be hidden even within the most advanced visualization approaches when ceramide abundances are plotted as a function of m/z, or the total carbon number in the molecule, versus its retention time (as shown in 16, 17). What happened to be hidden within all available data is the novel biological phenomenon in molecular organization of at least NS ceramides. As shown in Fig. 5, when plotting extracted MRM profiles separately for NS ceramide with C18-, C20-,and C22-sphingosines, it becomes clear that NS ceramides with the longest sphingoid base have the longest N-linked fatty acids. This suggests the unknown selectivity of ceramide synthases towards dihydrosphingosines with different chain length that is revealed only upon keratinocyte terminal differentiation and stratum corneum formation.

Figure 3. A survey scan to detect lipids with free hydroxy group (mostly ceramides) in human stratum corneum.

A neutral loss of 18Da scan performed with moderate declustering potential (80V) and low collision energy (20 eV) on lipid extract from human stratum corneum. This survey scan allows identifying major molecular ions for further structural analyses.

Figure 5. Preferred presence of very long-chain fatty acids in NS ceramides with the longest sphingoid base in human stratum corneum.

Multiple Reaction Monitoring (MRM) profiling of human stratum corneum NS ceramides demonstrates the increasing proportion of 26:0, 28:0, and 30:0-fatty acids within NS ceramides from C18- to C22-sphingosines. Arbitrary horizontal lines delineate approximate “slope” in overall fatty acid hydrophobicity within NS ceramides with different chain length of sphingoid base; vertical line connects NS ceramides with palmitic acid (N16:0). Ceramide detection was achieved in positive ions mode using AB Sciex 6500QTRAP mass spectrometer as a transition from molecular ions to the m/z 264 (S18), m/z 292 (S20), and m/z 320 (S22). HPLC separation was performed using Shimadzu Nexera-X2 UHPLC system on Ascentis Express RP-amide column (2.7 μm 2.1 × 50 mm) with gradient elution from methanol:water:formic acid (50:50:0.5, 5mM ammonium formate) to methanol:chloroform: water:formic acid (90:10:0.5:0.5, 5 mM ammonium formate).

STS processing protocol developed by our group [21,32,33] is unique in the capacity to obtain lipid and NMF data from the same sample and even to perform further metabolomic studies due to the removal of stratum corneum from strip adhesive before extraction. The example of simultaneous detection of cis-UCA, trans-UCA, PCA, and most of the amino acids is shown in Fig. 6 and follows in overall the separation protocol described in [48]. What has to be noted is that the gradient mixing of acetonitrile (Solvent A) and methanol:water:formic acid with ammonium formate (solvent B) has a profound effect on amino acid ionization efficiency that results in extreme assay sensitivity towards UCA and requires the use of standard curves of responses of variable amounts of UCA-PCA versus fixed amount of internal standard (in our case, U-[13C,15N]proline), while all other amino acids are quantified against corresponding stable isotope labeled analogs.

Conclusion

Skin tape stripping and the analysis of STS components by mass spectrometry have received increasing attention due to the non-invasiveness of sample collection, easy sample storage, and the power of mass spectrometric approaches for the analysis of hundreds and thousands of compounds. Numerous clinical studies are currently using STS approach to address the question of skin barrier dysfunction in allergic skin diseases and the ability to correct it through pharmacological interventions. Multiple complications make the task of STS processing challenging when compounds of different polarities are required to be analyzed from limited number of STS. The current review provides some guidance on achieving maximal results, points to unsolved questions in stratum corneum lipid analysis, and demonstrates benefits of stratum corneum separation from adhesives before processing for subsequent analyses. Further work is required to understand the role played by each individual ceramide subgroup for skin barrier function to select most characteristic species and to simplify stratum corneum lipidomics for the purposes of clinical studies. This work will transform current processing protocols into a clinical application to bring the power of modern analytical technologies to a bedside with immediate response capabilities.

Figure 4. Complexity of human stratum corneum ceramides.

(A) Product ions of the m/z 678.5 demonstrate that C18- and C20-sphingosines and phytosphingosines are the major sphingoid bases in ceramides with the molecular and pseudomolecular ion of the m/z 678.5. This corresponds to structural formulas CER(N26:0)S(18), CER(N24:0)S(20), CER(N26:0)P(18), and CER(N24:0)P(20). Inserts show the nomenclature of MS/MS fragments (left) and the presence of sphingoid bases with odd chain length (C19) (right) within ceramides with the m/z 678.5. (B) Precursor ions scan for the m/z 264.4 demonstrates the presence and the abundance of major NS and AS ceramide as wells as major EOS ceramide species that contain S(18) sphingoid base. Both experiments were performed with declustering potential of 80V and collision energy of 40eV on AB Sciex 6500QTRAP instrument.