The University of Manchester Research

Lipidomics for translational skin research: a primer for theuninitiatedDOI:10.1111/exd.13558

Document VersionAccepted author manuscript

Link to publication record in Manchester Research Explorer

Citation for published version (APA):Kendall, A. C., Koszyczarek, M. M., Jones, E. A., Hart, P. J., Towers, M., Griffiths, C. E. M., ... Nicolaou, A. (2018).Lipidomics for translational skin research: a primer for the uninitiated. Experimental Dermatology, 27(7), 721-728.https://doi.org/10.1111/exd.13558

Published in:Experimental Dermatology

Citing this paperPlease note that where the full-text provided on Manchester Research Explorer is the Author Accepted Manuscriptor Proof version this may differ from the final Published version. If citing, it is advised that you check and use thepublisher's definitive version.

General rightsCopyright and moral rights for the publications made accessible in the Research Explorer are retained by theauthors and/or other copyright owners and it is a condition of accessing publications that users recognise andabide by the legal requirements associated with these rights.

Takedown policyIf you believe that this document breaches copyright please refer to the University of Manchester’s TakedownProcedures [http://man.ac.uk/04Y6Bo] or contact uml.scholarlycommunications@manchester.ac.uk providingrelevant details, so we can investigate your claim.

Download date:18. Jan. 2020

1

Lipidomics for translational skin research: a primer for the

uninitiated

Alexandra C Kendall1, Marta M Koszyczarek1, Emrys A Jones2, Philippa J Hart2, Mark

Towers2, Christopher EM Griffiths3 Michael Morris2, Anna Nicolaou1*

1Laboratory for Lipidomics and Lipid Biology, Division of Pharmacy and Optometry, School

of Health Sciences, Faculty of Biology, Medicine and Health, University of Manchester,

Manchester Academic Health Science Centre, Manchester, M13 9PL, UK

2Waters Corporation, Stamford Avenue, Wilmslow, SK9 4AX, UK

3Dermatology Centre, Salford Royal Hospital, University of Manchester, Manchester

Academic Health Science Centre, Manchester, UK

*Corresponding Author: Professor Anna Nicolaou, Division of Pharmacy and Optometry,

School of Health Sciences, Faculty of Biology, Medicine and Health, University of

Manchester, Manchester M13 9PL, UK. Email: anna.nicolaou@manchester.ac.uk

Word count: 3773

Figure count: 2

Reference count: 99

Key words:

Skin lipids; mass spectrometry; mass spectrometry imaging; ceramides; chromatography

Abbreviations

3Q, triple quadrupole; CMC, carboxymethylcellulose; DESI, desorption electrospray

ionisation; ELISA, enzyme-linked immunosorbent assay(s); ESI, electrospray ionisation; FA,

fatty acid(s); MALDI, matrix-assisted laser desorption ionisation; MS, mass spectrometry;

QToF, quadrupole-time-of-flight; TEWL, trans-epidermal water loss; ToF, time-of-flight;

UHPSFC, ultra-high performance supercritical fluid chromatography; UPLC,

ultraperformance liquid chromatography; PUFA, polyunsaturated fatty acid(s).

2

Abstract

Healthy skin depends on a unique lipid profile to form a barrier that confers protection and

prevents excessive water loss, aids cell-cell communication, and regulates cutaneous

homeostasis and inflammation. Alterations in the cutaneous lipid profile can have severe

consequences for skin health and have been implicated in numerous inflammatory skin

conditions. Thus, skin lipidomics is increasingly of interest, and recent developments in mass

spectrometry-based analytical technologies can deliver in-depth investigation of cutaneous

lipids, providing insight into their role and mechanism of action. The choice of tissue

sampling technique and analytical approach depends on the location and chemistry of the

lipid of interest. Lipidomics can be conducted by various mass spectrometry approaches,

including different chromatography and ionisation techniques. Targeted mass spectrometry

is a sensitive approach for measuring low-abundance signaling lipids, such as eicosanoids,

endocannabinoids, and ceramides. This approach requires specific extraction,

chromatography and mass spectrometry protocols to quantitate the lipid targets. Untargeted

mass spectrometry reveals global changes and allows analysis of hundreds of complex

lipids across a range of lipid classes, including phospholipids, glycerophospholipids,

cholesteryl esters and sphingolipids. Mass spectrometry lipid imaging, including matrix-

assisted laser desorption ionisation mass spectrometry and desorption electrospray

ionisation mass spectrometry, can reveal information about abundance and anatomical

distribution of lipids within a single skin sample. Skin lipidomics can provide qualitative and

quantitative data on hundreds of biologically-relevant lipid species with different properties

and activities, all found within a single skin sample, and support translational studies

exploring the involvement of lipids in skin health and disease.

3

1. INTRODUCTION

The skin is a site of complex and unique lipid metabolism, with lipids essential to cutaneous

function and dependent on dietary provision [1, 2]. From fatty acyls to more complex

glycerolipids, glycerophospholipids, sphingolipids and sterols, the lipid classes found in skin

cover a range of chemistries, and perform niche roles within the organ [3, 4]. Sebum on the

skin’s surface predominantly comprises triacylglycerols, wax esters, squalene and free fatty

acids; lipid species that help to waterproof the skin, contribute to barrier function, but also

support and regulate cutaneous microbiota [5, 6]. An equimolar ratio of ceramides,

cholesterol and free fatty acids form the epidermal lipid barrier in the stratum corneum, and

are responsible for preventing excessive trans-epidermal water loss (TEWL) and ion

permeability [7]. The lipids found in skin cell membranes, predominantly

glycerophospholipids, sphingomyelins and cholesterol, play important roles in protein

function, cell mobility and cell-cell communications [8-10]. Importantly, membrane lipids also

undergo active metabolism to bioactive lipid mediators, including eicosanoids,

endocannabinoids and ceramides, which regulate numerous physiological processes in skin

homeostasis and inflammation, including immune cell mobility and function [11-13],

vasodilation and relaxation [14, 15], and keratinocyte and fibroblast proliferation,

differentiation and migration [16-18]. Cutaneous lipid metabolism results in a continuous flux

of bioactive lipid species, and the increasing appreciation of the translational role of skin

lipidomics has driven developments in lipidomic technologies. In order to obtain the

maximum amount of information relevant to skin lipid biology, it is important to use

bioanalytical approaches capable of identifying and quantitating across the range of lipids.

Here we present a review of the main technology platforms currently used in skin lipidomics,

and provide examples of the amount and variety of lipidomic information that can be

obtained from skin biopsies using these techniques, including targeted mass spectrometry,

untargeted mass spectrometry, and mass spectrometry imaging.

2. SKIN LIPID SAMPLING AND EXTRACTION

Skin lipid analysis requires appropriate lipid sampling, extraction and analysis, which depend

on the skin compartment of interest, the chemistry of the target lipids, and the information

required, respectively. Examples of different approaches are shown in Figure 1.

2.1. Skin lipid sampling

The choice of skin lipid sampling is determined by the location of the cellular target within the

skin. Surface lipids found in sebum or stratum corneum can be captured using non-invasive

tape strips. Tapes designed to absorb sebum lipids (e.g. Sebutape® [19]) or remove stratum

corneum (e.g. D-Squame® [20]) can be treated with solvents to yield epidermal lipids [21-

23]. Tape strips have been used to sample consecutive non-viable stratum corneum layers

and provide information on lipid species found at different stratum corneum depth [24], but

deeper sampling is needed to examine viable cells. If lipids from the dermis are also

required, the most appropriate sampling approach is the full-thickness skin biopsy. Biopsies

can be split to allow separate analyses of the different skin compartments, or used for lipid

imaging to analyse the full cross-section of the biopsy [25, 26]. A third option is suction

4

blister fluid sampling [27]. This involves applying vacuum to the skin to raise a suction blister,

followed by aspiration of the blister fluid that contains lipid mediators of primarily epidermal

origin [28]. Other types of samples that can be used to study cutaneous lipid biology include

the vernix caseosa [29] and sweat [30, 31]. Sample preparation is important, as rapid

degradation or metabolism of lipids is possible post tissue-harvesting. Ideally, skin samples

(biopsies, blister fluid or tape strips) should be flash-frozen in liquid nitrogen immediately

after harvesting. Samples should then be stored at -80 ºC prior to analysis.

2.2. Lipid extraction

Lipid extraction primarily relies on solvents to disrupt cellular membranes and solubilise

lipids. Solvents are selected based on the chemistry of the lipids being analysed, and vary

depending on the polarity of the target lipids (this has been reviewed recently [32]). Whilst

tape-stripped stratum corneum and epidermal blister fluid samples can be added directly to

solvents [20, 33], biopsy samples require homogenisation to disrupt the tissue and maximize

lipid extraction [34]. Some extraction protocols involve further purification of lipids, such as

solid-phase extraction, methylamine or alkaline hydrolysis, to remove contaminants and

reduce matrix effects that can impede the analysis; this is particularly important when

targeting lipid species found in low abundance [35-37]. Lipid imaging approaches (discussed

in Section 4) such as matrix-assisted laser desorption ionisation (MALDI) mass

spectrometry imaging (MSI) and desorption electrospray ionisation-MSI (DESI-MSI) require

no initial extraction steps, as lipids are ionised directly from the tissue sample.

The amount of lipid extracted can be normalised against tissue wet weight, dry weight,

sample volume, protein content, phosphate or cholesterol [28, 38, 39]. Most lipid extraction

protocols allow the retention of proteins from the sample, for later analysis of protein content.

However, consideration must be given to how cutaneous changes might influence the

sample protein content. For example, diseased skin may display different levels of cell

cohesion compared with healthy skin, resulting in the collection of a different number of cells

when using tape strips [40]. In these cases, it may be more appropriate to report total lipids

per surface area. Additionally, certain treatments may alter the protein content of blister fluid,

so normalization by fluid volume may be more appropriate [41].

3. SKIN LIPID ANALYSIS

Although approaches such as enzyme-linked immunosorbent assays (ELISA) can be used

to quantitate some lipid mediators (e.g. prostaglandins), there are limitations of these

methods. ELISA can only detect a single lipid species at a time, although are often unable to

differentiate between isomeric species, resulting in considerable cross-reactivity [42].

Additionally, discovery of new lipid species is impossible with ELISA, and limited lipid targets

are available. Some of the analytical approaches used in lipid analysis, including nuclear

magnetic resonance and gas chromatography, can offer some insight into skin lipid biology,

but there are limitations in their compound coverage, sensitivity and efficiency [43].

Developments in mass spectrometry and chromatography have accelerated lipidomic

analytical capabilities, allowing accurate identification and quantitation of complex mixtures

of lipids [44]. Mass spectrometry covers a range of analytical capabilities that can be

selected depending on the lipids of interest, relying on the identification of lipids based on

5

their ionisation and molecular fragmentation patterns. Although some lipid species are

analysed following untargeted shotgun approaches [20], quantitative lipidomics are

performed using mass spectrometry paired with chromatography to separate lipid

components prior to their analysis, to maximise sensitivity. Complex lipid mixtures found in

skin benefit from appropriate separation to improve the analytical sensitivity, minimise matrix

effects and ion suppression, and inform peak identification [45]. It is important to consider

both the quantity and proportion of lipids present in the skin. For example, aged skin appears

to contain lipids in the same proportions as young skin, but overall levels are reduced [46],

whereas atopic dermatitis skin demonstrates changes in both the overall level of cutaneous

ceramides, and the proportion of different ceramide families [47]. Such detailed

investigations require accurate lipid identification and quantitation, and a range of mass

spectrometry-based lipidomics approaches are available for consideration.

3.1 Chromatography

Chromatography is used to separate different lipid classes (or lipid species) found in a

sample following their extraction from tissue; the choice of chromatography depends on the

lipids of interest and their physico-chemical properties. Cutaneous lipids are a diverse family

of hydrophobic, low molecular weight compounds, with a range of polarities and

considerable structural similarity within each lipid class [48]. These aspects complicate both

lipid separation and analysis, and often different approaches will be required to fully

interrogate skin lipid biology.

Normal-phase chromatography separates complex lipids based on the functionality of the

polar head group, which results in similar retention times for members of the same lipid

class, making distinction between lipid isomers extremely difficult, although linear separation

by molecular weight is possible [49]. Normal phase chromatography also allows the rapid

elution of extremely hydrophobic compounds, such as acylceramides, which is not always

possible in reverse-phase chromatography [50]. Reverse-phase chromatography separates

lipids based on their fatty acyl components (different chain length and number of double

bonds), resulting in co-elution of different lipid classes, but allowing separation of isomers

within the same lipid class [51]. Supercritical fluid chromatography is a hybrid of normal- and

reverse-phase liquid chromatography, offering separation of molecules with a broad range of

polarities in a single chromatographic run [52]. Consequently, this is a useful approach for

the untargeted analysis of complex mixtures of skin lipids, including glycerolipids,

glycerophospholipids, sterols and free fatty acids, as hydrophilic, hydrophobic, and polar lipid

species can be separated based on both their fatty acyl moieties and their polar head

groups; this is useful for classes with similar hydrophobicities such as triacylglycerols and

cholesteryl esters, which are important in skin [52-56].

3.2. Mass spectrometry

Whether direct infusion or chromatographic introduction is used, there are various mass

spectrometry platforms available for lipidomics applications (recently reviewed in [44]).

Extracted lipids must be ionised and introduced into the mass spectrometer for analysis. The

preferred ionisation method for most lipids is electrospray ionisation (ESI). An electric

potential applied to a nebulised sample solution results in a fine spray, forming ions upon

desolvation [57, 58]. ESI is considered to be a softer ionisation technique than other popular

approaches, and provides a reliable tool for obtaining molecular mass information in

6

lipidomics [59]. Other common lipid ionisation methods include MALDI and DESI (discussed

in Section 4) [44]. DESI is a derivative of ESI, in which spatial distribution of lipids in the

sample can be maintained. Lipid ions can be analysed using different detection methods,

chosen based on the nature of the analyte and the kind of data required. Typically, tandem

mass spectrometry (MS/MS; a combination of two mass selection units) is used to monitor

both precursor ions and product ions resulting from precursor fragmentation in a collision

cell. Different linear combinations of quadrupoles (Q), time-of-flight (ToF) mass

spectrometers, and ion traps generate multiple options for lipidomics analysis (e.g. triple-

quadrupoles (3Q), QToFs and ToF-ToFs) [60]. These approaches can provide information

about both precursor and product ions, enabling more confident peak identification, and this

method of detection is particularly recommended for complex mixtures such as skin lipid

extracts.

3.2.1. Targeted mass spectrometry

Targeted mass spectrometry can be used to investigate a particular lipid family or pathway

of interest. Molecular fragmentation patterns can be identified for individual lipid species,

and, if the analysis includes separation by chromatography, retention times can provide

extra identifying information. Additionally, if lipid standards are available commercially or can

be synthesised, calibration lines can be constructed to allow accurate quantitation. A

discrete list of lipid species can thus be identified and quantitated from a complex lipid

extract. The most common approach for targeted quantitation is with multiple-reaction-

monitoring on 3Q mass spectrometers owing to the high degrees of sensitivity and

specificity.

This approach has been used to generate information about bioactive lipid mediators in skin,

including eicosanoids and related species, endocannabinoids and other N-acyl

ethanolamines [28, 34, 61-63]. As many of these pathways and lipid classes have been well

elucidated, individual species can be analysed using a targeted approach, and, by limiting

the range of lipids being analysed, sensitivity is maximised, which is important given that

these lipids are usually present in low abundance [42]. Another set of cutaneous lipids that

can be analysed using targeted lipidomics is the ceramide family, a group of lipids that are

essential components of the epidermal lipid barrier [64]. Although ceramides are found

throughout the body, the skin contains unique acylceramides with an additional linoleic acid

sub-group, and these are formed within the epidermis [65-67]. Ceramides have been

implicated in diseases of epidermal barrier dysfunction, including psoriasis [68], atopic

dermatitis [47] and ichthyosis [69], and so are of clinical interest. The analysis of ceramides

has progressed from simple, thin-layer chromatography, which provides total ceramide

content or analysis of the ceramide classes found in skin lipid extracts, including the complex

acylceramides, through qualitative identification of individual species, to accurate

identification and quantitation or semi-quantitation of more than 320 ceramide species using

multiple-reaction-monitoring by MS/MS [23, 47, 50, 64, 70-74].

Preliminary targeted analysis of epidermal and dermal ceramides performed by assays

developed in our group (Supplementary-S1) reveals differences in the profiles of the two

skin compartments (Fig. 2D). The most obvious difference is the absence of acylceramides

(CER[EOH], CER[EOS] and CER[EOP]) in the dermis, as these are only found in the

stratum corneum. Due to the limited availability of standards representing each ceramide

class, we performed relative semi-quantitation using a 43-carbon CER[NS] as internal

7

standard. However, this does not account for differences in ionisation efficiencies in different

lipid classes, and as class-specific deuterated standards, including standards representing

the very non-polar acylceramides, continue to become commercially available, ceramide

quantitation will become more accurate. Importantly, a similar targeted approach can be

used to analyse a greater range of bioactive lipid species with signalling roles in skin health

and disease, including eicosanoids and endocannabinoids [28, 34, 42]. However, this

analysis necessitates more sample as the extraction protocols of low-abundance species,

such as the eicosanoids, requires a different extraction protocol optimised for this lipid class

[28, 34]. Such low-abundance, short-lived lipid mediators are unlikely to be identified by

untargeted analyses or lipid imaging, so targeted mass spectrometry remains the most

appropriate platform for their analysis. Commercial availability of synthetic standards and

deuterated internal standards permits the use of calibration lines for quantitation, and the

resulting assays can reveal disease- or treatment-induced changes in specific lipid pathways

[33, 34, 73].

3.2.2. Untargeted mass spectrometry

Untargeted lipidomics can provide a useful and rapid approach to identify global changes in

lipid profiles, as well as discover novel lipid species. Although restrictions include limited

identifications due to the large number of lipid species being analysed, and reduced

sensitivity, hundreds of lipid species can be identified and semi-quantitated in one analytical

run [45, 75]. One of the simplest and fastest data acquisition modes for untargeted

lipidomics is MSE, which involves detection of molecules within a given mass-to-charge ratio

(m/z) range applying alternating low and high collision energies, using hybrid mass

spectrometers such as Q-ToF that provide accurate mass information to allow the

identification of co-eluting lipid analytes. In this approach, fragment ions are formed in an

untargeted manner, revealing precursor and product ions that can be matched by retention

time, enabling lipid identification. The approach is useful for lipids such as glycerolipids,

glycerophospholipids, and sterols. These classes of lipids are complex and contain multiple

fatty acids, with different combinations resulting in numerous different species. Some lipid

classes have characteristic fragmentation patterns (typically common head groups) which

aids identification if different lipid classes co-elute.

Untargeted UHPSFC/ESI-QToF analysis of human skin biopsies performed by our group

(Supplementary-S2) reveals a huge array of lipid species, including triacylglycerols,

diacylglycerols, cholesteryl esters, phospholipids including phosphatidylcholines and

phosphatidylethanolamines, and sphingomyelins. The most abundant species were

identified and semi-quantitated using class-specific deuterated internal standards, and

normalised against tissue protein content. Totals for each lipid class in this preliminary study

are shown in Fig. 2C, revealing differences in the lipid profiles between the dermis and

epidermis. This approach can identify global lipid changes in skin diseases, and may be

particularly useful in biomarker discovery.

4. MASS SPECTROMETRY LIPID IMAGING

The mass spectrometry approaches described above rely on the extraction of lipids from

tissue prior to analysis. However, this means that detailed anatomical information about the

8

location of lipids or lipid changes within skin is lost upon homogenisation of the tissue.

Another option for skin lipid analysis is mass spectrometry imaging (MSI). This allows the

direct sampling of lipid ions from the skin tissue, without prior extraction, so that lipids can be

analysed in situ [76]. While direct ionisation techniques such as MALDI and DESI can be

used for homogenised samples, they have also proved useful in providing spatial distribution

information through skin imaging analysis [26, 77-80]. A full-thickness skin biopsy can be

sectioned onto a glass slide, and by placing the sample on a moving stage, the mass

spectrometer samples ions as it performs spot-to-spot analysis, or rasters, across the skin

tissue, allowing reconstruction of an image of the full-thickness skin representing ion

intensities [81]. For each pixel with a given x,y coordinate, a mass spectrum is recorded with

information on m/z and ion abundance. By selecting an ion of interest and extracting the

data for each pixel, a chemical map of the tissue can be built up, showing the distribution

and abundance of that lipid molecular ion across the sample [82, 83]. This provides

information on the different skin compartments in one image, revealing anatomical lipid

distribution, as well as allowing semi-quantitative comparison of lipid intensities between

samples, and the MSI data can be co-registered with the same or a consecutive section that

has undergone histochemical staining [26, 77]. The size of the pixels and the ions sampled

depends on the ionisation approach, and MALDI and DESI offer different advantages.

4.1. MALDI

MALDI-MSI involves coating the tissue with a matrix that promotes ionisation of the lipid

species [84, 85], and has been used to study lipid changes in skin and skin equivalents [26,

77, 78]. MALDI-MSI can provide very good spatial resolution depending on the laser spot

size on the surface, with small pixel sizes of 15 µm obtainable using commercial systems

[86], but analysis requires special consideration of sample preparation, including matrix

application, which depends on factors such as the target ions (recently reviewed in [87]). An

important consideration of matrix selection is the spectrum of matrix-derived ions and

whether these interfere with the compounds of interest. Examples of matrices that are used

for lipid analysis include α-cyano-4-hydroxy-cinnamic acid (CHCA) and 9-aminoacridine

(9AA) [88]. CHCA typically produces ions below m/z 500 [89], so these do not interfere with

most lipid species of interest in skin, including phospholipids and acylglycerols [85]. 9AA is

commonly used in negative mode, and produces low m/z ions (majority below m/z 300) [90],

which would not interfere with lipids typically identified in negative mode, including

phosphatidylglycerols and phosphatidic acids.

4.2. DESI

DESI allows ionisation of lipids directly from skin tissue under ambient conditions by

directing an electrically-charged fine spray of an appropriate solvent at the tissue sample,

and sampling the ionised lipids directly into the mass spectrometer [91]. Unlike MALDI-MSI,

no matrix is required, so skin tissue simply requires sectioning prior to analysis. DESI-MSI

has proven capabilities in imaging of lipids across tissues [92-94], but until recently has

lacked the spatial resolution to analyse intricate tissues such as skin, with typical pixel sizes

around 100 µm [82, 83, 86].

4.3. Skin biopsy analysis

9

Untargeted lipid imaging analysis by MALDI-MSI (Supplementary-S3) and DESI-MSI

(Supplementary-S4) performed by our group provides examples of the images that can be

produced. MALDI-MSI was performed at a pixel size of 15 µm, and the image generated of

an example ion shows varied distribution across the tissue (the m/z 739.5 ion in positive

mode (Fig. 2A)), with increased intensity in the epidermis. DESI-MSI analysis was

performed at 40 µm pixel size (Fig. 2B), allowing analysis of the whole skin sections in the

same time-frame, although with less spatial resolution than MALDI-MSI. The image

generated shows the distribution and intensity of a different ion (m/z 789.5 in positive mode)

in three skin samples, again with increased abundance in the epidermis. Due to differences

in ionisation and adduct formation in the two techniques, different lipid ion profiles are often

identified, making direct comparison difficult. However, the two approaches may provide

complementary information on skin biology. These examples were chosen from untargeted

analysis because of clear differences across the tissue in this preliminary study, but targeted

approaches examining specific lipids of interest can also be performed using both MALDI-

MSI and DESI-MSI.

6. CONCLUSIONS AND FUTURE PERSPECTIVES

Lipidomics analysis of skin can provide a vast amount of information from limited tissue

samples, including bioactive lipid mediators, structural lipids, and lipid imaging. Lipid

imaging and untargeted analysis may not be able to detect low abundance short-lived

mediators such as eicosanoids, which are easily detectable by UPLC/ESI-MS/MS, whilst

targeted analysis may miss important changes in unknown lipids. To obtain a full

appreciation of the skin’s lipid profile, it may be necessary to combine several approaches.

Indeed, DESI-MSI and MALDI-MSI studies could be targeted to analyse a selection of lipids

identified through UPLC/ESI-MS/MS or UHPSFC/ESI-QToF, to investigate whether their

distribution throughout the skin changes as well as their abundance, whilst lipid ions

identified through lipid imaging could be followed up with more targeted analysis for

improved sensitivity. Lipid analysis can reveal diverse information about cutaneous lipid

biology, from mechanistic insight into inflammatory skin conditions such as sunburn [61, 62]

and irritant dermatitis [34], to the effects of topical treatments and nutritional supplementation

on the skin lipidome [33, 95-99], revealing biomarkers, unravelling complex signalling

pathways, and identifying potential therapeutic targets. As mass spectrometry technology

continues to improve, resulting in increased mass accuracy, sensitivity, spatial resolution

and more, we expect skin lipidomics to become a routine consideration in the investigation of

cutaneous homeostasis and inflammation.

10

Acknowledgements

This work was supported by the Medical Research Council (grant number MC_PC_15058).

We thank the NIHR Manchester Biomedical Research Centre for supporting this study: AN is

supported by the NIHR Manchester Biomedical Research Centre as a Dermatology Theme

Key Researcher; CEMG is an NIHR Senior Investigator. We thank Neil O’Hara (The

University of Manchester) for excellent technical support and Emmanuelle Claude (Waters

Corporation) for supporting the MALDI-MSI analysis. We acknowledge the support of the

National Institute of Health Research Clinical Research Network (NIHR CRN), and we would

like to thank Mr Jean Bastrilles and Mrs Marie Durkin for performing skin biopsies.

Author contributions

AN, MM and CEMG were responsible for the study design. MMK coordinated volunteer

recruitment and biopsy sampling and performed lipid analysis. ACK, EAJ, PJH and MT

performed lipid imaging analysis. AK analysed lipid data and drafted the manuscript. All

authors performed critical revision of the manuscript.

Conflicts of interest

EAJ, PJH, MT and MM are employees of Waters Corporation.

11

Figure legends

Figure 1. Schematic representation of skin lipidomics workflow. Several mass

spectrometry-based lipidomics approaches are available for the investigation of the skin

lipidome. A combination of these platforms can reveal the most information about skin lipids.

3Q, triple-quadrupole; IS, internal standards; RPC, reverse-phase chromatography;

UHPSFC, supercritical fluid chromatography; ESI, electrospray ionisation; MALDI, matrix-

assisted laser desorption ionisation; DESI, desorption electrospray ionisation.

Figure 2. Lipidomic analysis of skin by multiple approaches. 6 mm skin half-biopsies

(n=3) were embedded in carboxymethylcellulose and sectioned onto glass slides (10 µm

thickness sections) for MALDI-MSI (A) and DESI-MSI (B). Remaining unsectioned tissue

was washed, divided into dermis and epidermis, and analysed by UHPSFC/ESI-MS/MS for

structural lipids (C) and UPLC/ESI-MS/MS for ceramides (D). A) Matrix assisted laser

desorption ionisation mass spectrometry imaging (MALDI-MSI) of skin lipids. Skin

sections were analysed by MALDI-MSI at 15 µm pixel size. Representative image shows the

ion m/z 739.5 in positive mode, reconstructed by extraction of the ion from the mass

spectrum at each pixel. B) Desorption electrospray ionisation mass spectrometry

imaging (DESI-MSI) of skin lipids. Skin sections were analysed by DESI-MSI at 40 µm

pixel size. Representative image shows the ion m/z 789.5 in positive mode, reconstructed by

extraction of the ion from the mass spectrum at each pixel. C) Structural lipids in skin

samples. Structural lipids of the epidermis and dermis were analysed by UHPSFC/ESI-

MS/MS. The most abundant lipids were identified using in-house algorithms and lipid

databases, including LIPID MAPS, and semi-quantitated against class-specific internal

standards. Lipid data are shown as total µg/mg protein per lipid class, mean ± SD, n=3

(LPC, lysophosphatidylcholines; PE, phosphatidylethanolamines;, SM, sphingomyelins; PC,

phosphatidylcholines; CHL, cholesterol; CE, cholesteryl esters; DAG, diacylglycerols; TAG,

triacylglycerols). D) Ceramide analysis of skin samples. Ceramides of the epidermis and

dermis were analysed by UPLC/ESI-MS/MS and multiple-reaction-monitoring, and semi-

quantitated using internal standards. Ceramides are presented as total pmol/mg protein per

ceramide class, mean ± SD, n=3. Ceramide classes are named according to the

combination of fatty acid (N, non-hydroxy; A, alpha-hydroxy; EO, ester-linked omega-

hydroxy) and sphingoid base (S, sphingosine; DS, dihydrosphingosine; H, 6-

hydroxysphingosine; P, phytosphingosine).

Supplementary files

Supplementary-S1: Ceramide analysis

Supplementary-S2: Structural lipid analysis

Supplementary-S3: MALDI-MSI analysis

Supplementary-S4: DESI-MSI analysis

12

References

1. Burr G O, Burr M M. J Biol Chem 1929: 82: 345-367. 2. Burr G O, Burr M M. J Biol Chem 1930: 86: 587-621. 3. Fahy E, Subramaniam S, Murphy R C, Nishijima M, Raetz C R H, Shimizu T, Spener F, van Meer G, Wakelam M J O, Dennis E A. J Lipid Res 2009: 50: S9-S14. 4. Kendall A C, Nicolaou A. Prog Lipid Res 2013: 52: 141-164. 5. Picardo M, Ottaviani M, Camera E, Mastrofrancesco A. Dermatoendocrinol 2009: 1: 68-71. 6. Pappas A. Dermatoendocrinol 2009: 1: 72-76. 7. Elias P M. J Invest Dermatol 2005: 125: 183-200. 8. Sunshine H, Iruela-Arispe M L. Curr Opin Lipidol 2017: 28: 408-413. 9. Head B P, Patel H H, Insel P A. Biochim Biophys Acta 2014: 1838: 532-545. 10. van Meer G, Voelker D R, Feigenson G W. Nat Rev Mol Cell Biol 2008: 9: 112-124. 11. Lone A M, Taskén K. Front Immunol 2013: 4: 130. 12. Nicolaou A, Mauro C, Urquhart P, Marelli-Berg F. Front Immunol 2014: 5: 75. 13. Poloso N J, Urquhart P, Nicolaou A, Wang J, Woodward D F. Mol Immunol 2013: 54: 284-295. 14. Félétou M, Huang Y, Vanhoutte P M. Br J Pharmacol 2011: 164: 894-912. 15. Mair K M, Robinson E, Kane K A, Pyne S, Brett R R, Pyne N J, Kennedy S. Br J Pharmacol 2010: 161: 176-192. 16. Liu M, Saeki K, Matsunobu T, Okuno T, Koga T, Sugimoto Y, Yokoyama C, Nakamizo S, Kabashima K, Narumiya S, Shimizu T, Yokomizo T. J Exp Med 2014: 211: 1063-1078. 17. Wijesinghe D S, Brentnall M, Mietla J A, Hoeferlin L A, Diegelmann R F, Boise L H, Chalfant C E. J Lipid Res 2014: 55: 1298-1309. 18. Paradisi A, Pasquariello N, Barcaroli D, Maccarrone M. J Biol Chem 2008: 283: 6005-6012. 19. Camera E, Ludovici M, Galante M, Sinagra J-L, Picardo M. J Lipid Res 2010: 51: 3377-3388. 20. Sadowski T, Klose C, Gerl M J, Wójcik-Maciejewicz A, Herzog R, Simons K, Reich A, Surma M A. Sci Rep 2017: 7: 43761. 21. Ashraf Z, Pasha U, Greenstone V, Akbar J, Apenbrinck E, Foulks G N, Borchman D. Curr Eye Res 2011: 36: 553-562. 22. Starr N J, Johnson D J, Wibawa J, Marlow I, Bell M, Barrett D A, Scurr D J. Anal Chem 2016: 88: 4400-4408. 23. van Smeden J, Hoppel L, van der Heijden R, Hankemeier T, Vreeken R J, Bouwstra J A. J Lipid Res 2011: 52: 1211-1221. 24. Weerheim A, Ponec M. Arch Dermatol Res 2001: 293: 191-199. 25. Iuliana P, Jacques P. J Clin Exp Dermatol Res 2015: 6: 282-285. 26. Hart P J, Francese S, Claude E, Woodroofe M N, Clench M R. Anal Bioanal Chem 2011: 401: 115-125. 27. Vermeer B J, Reman F C, van Gent C M. J Invest Dermatol 1979: 73: 303-305. 28. Kendall A C, Pilkington S M, Massey K A, Sassano G, Khodes L E, Nicolaou A. J Invest Dermatol 2015: 135: 1510-1520. 29. Checa A, Holm T, Sjödin M O D, Reinke S N, Alm J, Scheynius A, Wheelock C E. Sci Rep 2015: 5: 15740. 30. Takemura T, Wertz P W, Sato K. Br J Dermatol 1989: 120: 43-47. 31. Agrawal K, Hassoun L A, Foolad N, Pedersen T L, Sivamani R K, Newman J W. J Lipid Res 2016. 32. Pati S, Nie B, Arnold R D, Cummings B S. Biomed Chromatogr 2016: 30: 695-709. 33. Pilkington S M, Gibbs N K, Costello P, Bennett S P, Massey K A, Friedmann P S, Nicolaou A, Rhodes L E. Exp Dermatol 2016: 25: 962-968. 34. Kendall A C, Pilkington S M, Sassano G, Rhodes L E, Nicolaou A. Br J Dermatol 2016: 175: 163-171. 35. Massey K A, Nicolaou A. Free Radic Biol Med 2013: 59: 45-55. 36. Wu J-x, Wu J-l, Yin J, Zheng P, Yao N. Front Plant Sci 2015: 6.

13

37. Singh A, Del Poeta M. Front Microbiol 2016: 7. 38. Senkal C E, Salama M F, Snider A J, Allopenna J J, Rana N A, Koller A, Hannun Y A, Obeid L M. Cell Metab 2017: 25: 686-697. 39. Joo K-M, Hwang J-H, Bae S, Nahm D-H, Park H-S, Ye Y-M, Lim K-M. J Dermatol Sci 2015: 77: 71-74. 40. Clausen M-L, Slotved H C, Krogfelt K A, Agner T. Scientific Reports 2016: 6: 19918. 41. Paquet P, Meuwis M A, Mazzucchelli G, Delvenne P, Pierard G E. Curr Drug Saf 2012: 7: 339-351. 42. Astarita G, Kendall A C, Dennis E A, Nicolaou A. Biochim Biophys Acta 2015: 1851: 456-468. 43. Gross R W, Han X. Chem Biol 2011: 18: 284-291. 44. Yang K, Han X. Trends Biochem Sci 2016: 41: 954-969. 45. Lísa M, Cífková E, Khalikova M, Ovčačíková M, Holčapek M. J Chromatogr A 2017. 46. Ghadially R, Brown B E, Sequeira-Martin S M, Feingold K R, Elias P M. J Clin Invest 1995: 95: 2281-2290. 47. Ishikawa J, Narita H, Kondo N, Hotta M, Takagi Y, Masukawa Y, Kitahara T, Takema Y, Koyano S, Yamazaki S, Hatamochi A. J Invest Dermatol 2010: 130: 2511-2514. 48. Scherer M, Bottcher A, Schmitz G, Liebisch G. Biochim Biophys Acta 2011: 1811: 68-75. 49. Zhu C, Dane A, Spijksma G, Wang M, van der Greef J, Luo G, Hankemeier T, Vreeken R J. J Chromatogr A 2012: 1220: 26-34. 50. Masukawa Y, Narita H, Shimizu E, Kondo N, Sugai Y, Oba T, Homma R, Ishikawa J, Takagi Y, Kitahara T, Takema Y, Kita K. J Lipid Res 2008: 49: 1466-1476. 51. Nakanishi H, Iida Y, Shimizu T, Taguchi R. J Biochem 2010: 147: 245-256. 52. Lisa M, Holcapek M. Anal Chem 2015: 87: 7187-7195. 53. Aschrafi A, Franzen R, Shabahang S, Fabro D, Pfeilschifter J, Huwiler A. Biochim Biophys Acta 2003: 1634: 30-39. 54. Bamba T, Shimonishi N, Matsubara A, Hirata K, Nakazawa Y, Kobayashi A, Fukusaki E. J Biosci Bioeng 2008: 105: 460-469. 55. Yamada T, Uchikata T, Sakamoto S, Yokoi Y, Nishiumi S, Yoshida M, Fukusaki E, Bamba T. J Chromatogr A 2013: 1301: 237-242. 56. Hutchins P M, Barkley R M, Murphy R C. J Lipid Res 2008: 49: 804-813. 57. Yamashita M, Fenn J B. J Phys Chem 1984: 88: 4671-4675. 58. Ho C S, Lam C W K, Chan M H M, Cheung R C K, Law L K, Lit L C W, Ng K F, Suen M W M, Tai H L. Clin Biochem Rev 2003: 24. 59. Postle A D, Wilton D C, Hunt A N, Attard G S. Prog Lipid Res 2007: 46: 200-224. 60. Köfeler H C, Fauland A, Rechberger G N, Trötzmüller M. Metabolites 2012: 2: 19-38. 61. Nicolaou A, Masoodi M, Gledhill K, Haylett A K, Thody A J, Tobin D J, Rhodes L E. Photochem Photobiol Sci 2012. 62. Rhodes L E, Gledhill K, Masoodi M, Haylett A K, Brownrigg M, Thody A J, Tobin D J, Nicolaou A. FASEB J 2009: 23: 3947-3956. 63. Sasso O, Pontis S, Armirotti A, Cardinali G, Kovacs D, Migliore M, Summa M, Moreno-Sanz G, Picardo M, Piomelli D. PNAS 2016: 113: E4397-E4406. 64. van Smeden J, Janssens M, Gooris G S, Bouwstra J A. Biochim Biophys Acta 2014: 1841: 295-313. 65. Jennemann R, Rabionet M, Gorgas K, Epstein S, Dalpke A, Rothermel U, Bayerle A, van der Hoeven F, Imgrund S, Kirsch J, Nickel W, Willecke K, Riezman H, Gröne H-J, Sandhoff R. Hum Mol Genet 2012: 21: 586-608. 66. Rabionet M, Bayerle A, Marsching C, Jennemann R, Gröne H-J, Yildiz Y, Wachten D, Shaw W, Shayman J A, Sandhoff R. J Lipid Res 2013: 54: 3312-3321. 67. Macheleidt O, Sandhoff K, Kaiser H W. J Invest Dermatol 2002: 119: 166-173. 68. Lew B-L, Cho Y, Kim J, Sim W-Y, Kim N-I. J Korean Med Sci 2006: 21: 95-99.

14

69. Eckl K-M, Tidhar R, Thiele H, Oji V, Hausser I, Brodesser S, Preil M-L, Önal-Akan A, Stock F, Müller D, Becker K, Casper R, Nürnberg G, Altmüller J, Nürnberg P, Traupe H, Futerman A H, Hennies H C. J Invest Dermatol 2013: 133: 2202-2211. 70. Kendall A C, Kiezel-Tsugunova M, Brownbridge L C, Harwood J L, Nicolaou A. Biochim Biophys Acta 2017. 71. Ishikawa J, Shimotoyodome Y, Ito S, Miyauchi Y, Fujimura T, Kitahara T, Hase T. Arch Dermatol Res 2012: 305: 151-162. 72. Tamura E, Ishikawa J, Naoe A, Yamamoto T. Int J Cosmet Sci 2016: 38: 615-621. 73. Pappas A, Kendall A C, Brownbridge L C, Batchvarova N, Nicolaou A. Exp Dermatol 2018: n/a-n/a. 74. Berkers T, van Dijk L, Absalah S, van Smeden J, Bouwstra J A. Exp Dermatol 2017: 26: 36-43. 75. Lísa M, Holčapek M. Anal Chem 2015: 87: 7187-7195. 76. Vickerman J C. Analyst 2011: 136: 2199-2217. 77. Mitchell C A, Donaldson M, Francese S, Clench M R. Methods 2016: 104: 93-100. 78. Mitchell C A, Long H, Donaldson M, Francese S, Clench M R. Lipids Health Dis 2015: 14: 84. 79. Goto-Inoue N, Hayasaka T, Zaima N, Nakajima K, Holleran W M, Sano S, Uchida Y, Setou M. PLoS ONE 2012: 7: e49519. 80. Motoyama A, Kihara K. Mass Spectrom (Tokyo) 2017: 6: S0071. 81. Taverna D, Pollins A C, Nanney L B, Sindona G, Caprioli R M. Exp Dermatol 2016: 25: 143-146. 82. Tillner J, McKenzie J S, Jones E A, Speller A V M, Walsh J L, Veselkov K A, Bunch J, Takats Z, Gilmore I S. Anal Chem 2016: 88: 4808-4816. 83. Tillner J, Wu V, Jones E A, Pringle S D, Karancsi T, Dannhorn A, Veselkov K, McKenzie J S, Takats Z. J Am Soc Mass Spectrom 2017: 28: 2090-2098. 84. Caldwell R L, Caprioli R M. Mol Cell Proteomics 2005: 4: 394-401. 85. Calvano C D, Carulli S, Palmisano F. Rapid Commun Mass Spectrom 2009: 23: 1659-1668. 86. Nielsen M M B, Lambertsen K L, Clausen B H, Meyer M, Bhandari D R, Larsen S T, Poulsen S S, Spengler B, Janfelt C, Hansen H S. Sci Rep 2016: 6: 39571. 87. de Macedo C S, Anderson D M, Schey K L. Talanta 2017: 174: 325-335. 88. Fuchs B, Süß R, Schiller J. Prog Lipid Res 2010: 49: 450-475. 89. Keller B O, Li L. J Am Soc Mass Spectrom 2000: 11: 88-93. 90. Vermillion-Salsbury R L, Hercules D M. Rapid Commun Mass Spectrom 2002: 16: 1575-1581. 91. Takáts Z, Wiseman J M, Gologan B, Cooks R G. Science 2004: 306: 471-473. 92. Abbassi-Ghadi N, Golf O, Kumar S, Antonowicz S, McKenzie J S, Huang J, Strittmatter N, Kudo H, Jones E A, Veselkov K, Goldin R, Takats Z, Hanna G B. Cancer Res 2016: 76: 5647. 93. Škrášková K, Claude E, Jones E A, Towers M, Ellis S R, Heeren R M A. Methods 2016: 104: 69-78. 94. Abbassi-Ghadi N, Jones E A, Gomez-Romero M, Golf O, Kumar S, Huang J, Kudo H, Goldin R D, Hanna G B, Takats Z. J Am Soc Mass Spectrom 2016: 27: 255-264. 95. Pilkington S M, Murphy S A, Kudva S, Nicolaou A, Rhodes L E. Exp Dermatol 2015: 24: 790-791. 96. Farrar M D, Nicolaou A, Clarke K A, Mason S, Massey K A, Dew T P, Watson R E B, Williamson G, Rhodes L E. Am J Clin Nutr 2015: 102: 608-615. 97. Pilkington S M, Rhodes L E, Al-Aasswad N M I, Massey K A, Nicolaou A. Mol Nutr Food Res 2014: 58: 580-590. 98. Rhodes L E, Darby G, Massey K A, Clarke K A, Dew T P, Farrar M D, Bennett S, Watson R E B, Williamson G, Nicolaou A. Br J Nutr 2013: 110: 891-900. 99. McDaniel J C, Massey K, Nicolaou A. Wound Repair Regen 2011: 19: 189-200.

Figure 1.

Figure 2.

Supplementary-S1 – ceramide analysis

Materials

Extraction solvents were obtained from Fisher Scientific (Loughborough, UK). LC/MS-grade

chromatography solvents were obtained from Sigma (Gillingham, UK).Ceramide/Sphingoid

Internal Standard Mixture I was obtained from Avanti Polar Lipids (Alabaster, USA).

Sample preparation

Skin biopsies (6 mm diameter) were obtained with informed written consent from volunteers

at The Salford Royal Hospital with full ethical approval from a local ethics committee and

according to the Declaration of Helsinki. Punch biopsies (6 mm) were cut, bisected, snap-

frozen in liquid nitrogen, and stored at -80°C prior to analysis. Dermis and epidermis of half-

biopsies were separated by scalpel and lipid extraction was performed according to the

modified Folch lipid extraction procedure[1]. 3 mL ice-cold 2:1 (v/v) chloroform:methanol was

added to the sample with 50 pmol ceramide internal standard cocktail (along with 50 µL

deuterated lipid cocktail for glycerolipid/glycerophospholipid analysis, described in

Supplementary-S3). Tissue was homogenised using a blade homogeniser (X 10/25 drive

with 10 mm diameter shaft, set at a speed of 11 kHz; Ystral, Ballrechten-Dottingen,

Germany) on ice [2, 3]. Homogenates were incubated on ice for 90 min to allow solvent lipid

extraction. Next, 500 µL of ice-cold, purified water was added to the vial and centrifuged.

The organic layer was removed, separated into two equal aliquots (one for

glycerolipid/glycerophospholipid analysis) and the ceramide aliquot was dried under

nitrogen, before reconstitution in 150 µL methanol with 0.1 % formic acid and storage at -20

ºC until analysis. The protein layer (above the organic layer) was kept for analysis of protein

content.

UPLC/ESI-MS/MS conditions

Analytes were separated using reverse-phase chromatography on a C8 column (Acquity

UPLC BEH, 1.7 µm, 2.1 x 100 mm; Waters, Wilmslow, UK) using a gradient of water (with

0.1% formic acid) and methanol (with 0.1% formic acid) as previously described [4].

Ceramides were identified on a Xevo TQ-S (Waters, Wilmslow, UK) using multiple reaction

monitoring in the positive ion mode (Fig S1-3). Since standards are not yet commercially

available for each individual ceramide compound, relative quantitation of ceramides was

performed using internal standards.

Protein content analysis

The protein content of the skin samples was analysed using Bio-Rad Protein Assay II (Bio-

Rad, Hemel Hempstead, UK).

Figure S1. Representative chromatograms of CER[EOH] ceramides in healthy

epidermis. Acylceramides from the CER[EOH] family were extracted from healthy epidermis

samples and analysed by UPLC/ESI-MS/MS. Individual ceramide species were identified by

their unique fragmentation pattern and retention time.

Figure S2. Representative chromatograms of CER[EOP] ceramides in healthy

epidermis. Acylceramides from the CER[EOP] family were extracted from healthy epidermis

samples and analysed by UPLC/ESI-MS/MS. Individual ceramide species were identified by

their unique fragmentation pattern and retention time.

Figure S3. Representative chromatograms of CER[EOS] ceramides in healthy

epidermis. Acylceramides from the CER[EOS] family were extracted from healthy epidermis

samples and analysed by UPLC/ESI-MS/MS. Individual ceramide species were identified by

their unique fragmentation pattern and retention time.

References

1. Folch J, Lees M, Stanley G H S. J Biol Chem 1957: 226: 497-509. 2. Kendall A C, Pilkington S M, Massey K A, Sassano G, Khodes L E, Nicolaou A. J Invest Dermatol 2015: 135: 1510-1520. 3. Kendall A C, Pilkington S M, Sassano G, Rhodes L E, Nicolaou A. Br J Dermatol 2016: 175: 163-171. 4. Kendall A C, Kiezel-Tsugunova M, Brownbridge L C, Harwood J L, Nicolaou A. Biochim Biophys Acta 2017.

1

Supplementary S2 – structural lipid analysis

Materials

Methanol, ethanol, 2-propanol, chloroform (all HPLC grade), and ammonium acetate were

purchased from Sigma Aldrich (Gillingham, UK). Deionised water was prepared with Elga

Water Purelab Flex purification system. Liquid CO2 CP grade (99.995%) was supplied by

BOC (Guildford, UK).

Deuterated lipid standards representative of 12 lipid classes (CE, cholesteryl esters; Chl,

cholesterol; SM, sphingomyelins; PC, phosphatidylcholines; LPC, lysophosphatidylcholines;

PE, phosphatidylethanolamines; TAG, triacylgycerols; DAG, diacylglycerols; PG,

phosphatidylglycerols; PA, phosphatidic acids; LPE, lysophosphatidylethanolamines; FFA,

free fatty acids) were used for UHPSFC/ESI-MS development, lipid identification and semi-

quantitation: 15:0 cholesteryl (d7) ester (CE-d7), cholesterol-d7 (Chl-d7), 16:0-d31

sphingomyelin (SM-d31), 16:0-d31-18:1 phosphatidylcholine (PC-d31), 26:0-d4

lysophosphatidylcholine (LPC-d4), 16:0-d31-phosphatidtylethanolamine (PE-d31), 17:0-17:1-

17:0-d5 triacylglycerol (TAG-d5), 18:1-d5 diacylglycerol (DAG-d5), 16:0-d31-18:1

phosphatidylglycerol (PG-d31), 16:0-d31-18:1 phosphatidic acid (PA-d31), 18:1-d7

lysophosphatidylethanolamine (LPE-d7), and palmitic acid-d31 (FFA-d31).

A deuterated lipid cocktail was prepared using individual deuterated lipid standards at final

concentrations comparable to the most abundant species of that class found in skin. Final

concentrations in the cocktail were: 100 µg/mL (FFA-d31, CE-d7, PC-d31, PE-d31, TAG-d5,

LPC-d4, SM-d31, DAG-d5, LPE-d7); 200 µg/mL (Chl-d7); 300 µg/mL (PG-d31, PA-d31).

Sample preparation

Skin biopsies (6 mm diameter) were obtained with informed written consent from volunteers

at The Salford Royal Hospital with full ethical approval from a local ethics committee and

according to the Declaration of Helsinki. Punch biopsies (6 mm) were cut, bisected, snap-

frozen in liquid nitrogen, and stored at -80°C prior to analysis. Dermis and epidermis of half-

biopsies were separated by scalpel and extraction was performed according to the modified

Folch lipid extraction procedure [1]. 3 mL ice-cold 2:1 (v/v) chloroform:methanol was added

to the sample with 50 µL deuterated lipid cocktail (along with 50 pmol Ceramide/Sphingoid

Internal Standard Mixture I, Avanti Polar Lipids, Alabaster, USA for ceramide analysis,

described in Supplementary-S1). Tissue was homogenised using a blade homogeniser (X

10/25 drive with 10 mm diameter shaft, set at a speed of 11 kHz; Ystral, Ballrechten-

Dottingen, Germany) on ice [2, 3]. Homogenates were incubated on ice for 90 min to allow

solvent lipid extraction. Next, 500 µL of ice-cold, purified water was added to the vial and

centrifuged. The organic layer was removed, separated into two equal aliquots (one for

ceramide analysis) and the glycerolipid/glycerophospholipid aliquot was dried under

nitrogen. The lipid extract was reconstituted in 1 mL 1:1 (v/v) chloroform-2-propanol and

stored at -20 ºC until analysis.

UHPSFC/ESI-MS conditions

Lipid analytes were separated by ultra-high performance supercritical fluid chromatography

(UHPSFC) [4] on an Acquity UPC2 instrument (Waters, Wilmslow, UK) using an Acquity

2

UPC2 Torus 2PIC column (100mm x 3mm x 1/7µm, Waters, Wilmslow, UK) under the

conditions outlined in Table S1 and using the gradient described in Table S2.

Table S1: UHPSFC conditions

Flow rate 1.5 µL/min

Injection volume (positive mode) 2 µL

Injection volume (negative mode) 7 µL

Column temperature 60 ºC

Active back pressure regulator 1800 psi

Injector needle wash 100 % methanol

Mobile phase A 100 % liquid CO2

Mobile phase B 99:1 (v/v) methanol:water with 30

mM ammonium acetate

Make-up solvent 99:1 (v/v) methanol:water with 30

mM ammonium acetate

Make-up solvent flow-rate 0.25 µL/min

Table S2: UHPSFC gradient

Time (min) Mobile phase A (%) Mobile phase B (%)

0 99 1

0.1 99 1

0.5 90 10

1.5 90 10

8 49 51

9 49 51

10 99 1

11 99 1

The UHPSFC instrument was coupled to a Synapt G2 High Definition Q-ToF Mass

Spectrometer (Waters, Milford, MA, USA) with electrospray ionisation. The mass

spectrometer was used in high-sensitivity mode according to conditions outlined in Table S3

in positive mode (CE, Chl, TAG, DAG, PC, LPC, SM, PE, LPE) or negative mode (PG, PA,

FA).

Table S3: Mass spectrometer conditions

Mass range m/z 50-1100

Capillary voltage 2.5 kV

Sampling cone 20 V

Source offset 90 V

Source temperature 150 ºC

Drying temperature 500 ºC

Cone gas flow 0.8 L/min

Drying gas flow 1.7 L/min

Nebuliser gas flow 4 bar

Lock-spray Leucine enkephalin (m/z 556.6)

Calibration reference Sodium formate

Acquisition mode MSE

Low collision energy 4 V

High collision energy 15-30 V ramp

3

Method Validation

The reproducibility of peak area was determined from 10 consecutive measurements of lipid

cocktail. The reproducibility of retention times was calculated as a standard deviation from

10 consecutive injections of the standards.

Lipid identification and nomenclature

Data processing was carried out using a data processing algorithm. Lipid identification of the

most abundant species was performed and confirmed through fragmentation patterns, using

the LIPID MAPS database and in-house information and publications. Representative

spectra of different lipid classes are shown in Fig. S1-5, including identification of the most

abundant species. The total number of species identified in each lipid class is shown in

Table S4.

Table S4. Number of lipid species of each class identified in human dermis and epidermis

Lipid class Epidermis Dermis

Cholesteryl esters 9 7

Triacylglycerols 83 84

Diacylglycerols 49 52

Phosphatidylcholines 16 17

Sphingomyelins 5 7

Phosphatidylethanolamines 2 4

Semi-quantitative analysis

Lipids were semi-quantitated based on mass spectra peak areas and those of the lipid class

standards. Following tissue homogenisation and extraction of the lipids from the biopsies,

the protein precipitate was collected and the protein content measured using Bio-Rad

Protein Assay II (Bio-Rad, Hemel Hempstead, UK). Lipid concentrations were calculated in

µg/mg of protein.

4

Figure S1. Representative spectra of triacylglycerols in epidermis and dermis. Individual triacylglycerol (TAG) species were identified

through fragmentation patterns, using the LIPID MAPS database and in-house information and publications. Epidermis and dermis show

differences in terms of ions present and relative abundances.

5

Figure S2. Representative spectra of diacylglycerols in epidermis and dermis. Individual diacylglycerol (DAG) species were identified

through fragmentation patterns, using the LIPID MAPS database and in-house information and publications. Epidermis and dermis show

differences in terms of ions present and relative abundances.

6

Figure S3. Representative spectra of phospholipids in epidermis and dermis. Individual phospholipid species including sphingomyelins (SM)

and phosphatidylcholines (PC) were identified through fragmentation patterns, using the LIPID MAPS database and in-house information and

publications. Epidermis and dermis show differences in terms of ions present and relative abundances.

7

Figure S4. Representative spectra of cholesteryl esters in epidermis and dermis. Individual cholesteryl ester (CE) species were identified

through fragmentation patterns, using the LIPID MAPS database and in-house information and publications. Epidermis and dermis show

differences in terms of ions present and relative abundances.

8

Figure S5. Representative spectra of free fatty acids in epidermis and dermis. Individual free fatty acid (FFA) species were identified through

fragmentation patterns, using the LIPID MAPS database and in-house information and publications. Epidermis and dermis show differences in

terms of ions present and relative abundances.

9

Supplementary references

1. Folch J, Lees M, Stanley G H S. J Biol Chem 1957: 226: 497-509. 2. Kendall A C, Pilkington S M, Massey K A, Sassano G, Khodes L E, Nicolaou A. J Invest Dermatol 2015: 135: 1510-1520. 3. Kendall A C, Pilkington S M, Sassano G, Rhodes L E, Nicolaou A. Br J Dermatol 2016: 175: 163-171. 4. Lísa M, Holčapek M. Anal Chem 2015: 87: 7187-7195.

1

Supplementary-S3 – Matrix assisted laser desorption ionisation mass

spectrometry imaging (MALDI-MSI)

Materials

All organic solvents, MALDI matrices and associated additives were supplied by Sigma

Aldrich (Dorset, UK). All organic solvents were of HPLC grade. Deionised water was

prepared with Elga Water Purelab Flex purification system.

Sample preparation

Skin biopsies (6 mm diameter) were obtained with informed written consent from volunteers

at The Salford Royal Hospital with full ethical approval from a local ethics committee and

according to the Declaration of Helsinki. Punch biopsies (6 mm) were cut, bisected, snap-

frozen in liquid nitrogen, and stored at -80°C prior to analysis. Half-biopsies were embedded

in 2 % (v/v) carboxymethylcellulose (CMC), sectioned (10 µm thick) onto glass slides,, and

stored at -80°C prior to analysis. For analyses in positive ionisation mode, ultra-pure α-

Cyano-4-hydroxycinnamic acid (α-CHCA) was used as the MALDI matrix. Here, the matrix

was dissolved in a solution of 70% ethanol, 30% water, 0.1% TFA, to a final concentration of

5 mg/mL. An equimolar amount of aniline was then added to the solution before a 5 minute

sonication step. The aniline is used to limit the adduction of CHCA, reducing the related

signals observed in the mass spectra and as such improving the visualisation of lipid related

ions [1]. For analyses in negative ionisation mode, 9-aminoacridine (9AA) was used as the

MALDI matrix, and was dissolved in a solution of 80% ethanol, 30% water. In every case,

matrix solutions were sprayed onto the sectioned skin using a SuncollectTM (SunChrom,

Friedrichsdorf, Germany), automated spraying device to give a homogenous coating.

MALDI-MSI conditions

All imaging acquisitions were set up in High Definition Imaging (HDI) informatics, version 1.4.

Slides were scanned using a standard flat-bed scanner (Epson). Images were acquired in

both positive and negative ionisation modes on a Synapt G2-Si (Q-TOF-MS) (Waters,

Wilmslow, UK).

A modified MALDI source using a custom acquisition script to allow higher acquisition rates

was utilised. An Nd:YAG solid state laser was used with a firing rate of 1000Hz. Data were

acquired at approximately 20 pixels per second in a continuous raster mode. The custom

script utilised the Waters Research Enabled Software (WREnS) platform to allow control of

the source, bypassing elements of the standard instrument control. The modified MALDI

source accommodates a larger plate capacity with positions for up to 3 slides. This allowed

for all samples in this experiment to be loaded and acquisitions queued and initiated without

further user intervention. Additionally, the laser geometry of this source has been adjusted to

allow for a laser focus of between 15-20 µm. Consequently, this allowed for MALDI-MS

images to be acquired at a spatial resolution (and final pixel size) of 15 µm.

Supplementary references

1. Calvano C D, Carulli S, Palmisano F. Rapid Commun Mass Spectrom 2009: 23: 1659-1668.

1

Supplementary-S4 – Desorption electrospray ionisation mass spectrometry

imaging (DESI-MSI)

Materials

All organic solvents and water used were of HPLC grade and supplied by Sigma Aldrich

(Dorset, UK). Any additives used in the investigation of optimum conditions were also

supplied by Sigma Aldrich and the solutions prepared fresh on the day of analysis.

Sample preparation and sampling optimisation

As discussed in the main article, the DESI-MSI technique requires no specific preparation of

the tissue sections prior to analysis. Fresh frozen sections of the tissue are collected at a

sample dependent thickness between 10 and 30 µm (typically 12 µm) and stored at -80°C

until required. Prior to analysis the samples are warmed to room temperature, if the samples

have a high aqueous content this can be done in a vacuum desiccator, and then placed onto

the sampling stage of the DESI source.

Skin biopsies (6 mm diameter) were obtained with informed written consent from volunteers

at The Salford Royal Hospital with full ethical approval from a local ethics committee and

according to the Declaration of Helsinki. Punch biopsies (6 mm) were cut, bisected, snap-

frozen in liquid nitrogen, and stored at -80°C prior to analysis. Half-biopsies were embedded

in 2 % (v/v) carboxymethylcellulose (CMC), sectioned (10 µm thick) onto glass slides, and

stored at -80°C prior to analysis.

DESI-MSI

The DESI-MSI process involves a focussed spray of electrostatically charged droplets

impacting upon a surface. The composition of the solvent used is selected to favour the

process of dissolving and ionising the molecular species of interest. For the lipid analysis

described here, high methanol content such as 98:2 (v/v) methanol:water was found to be

most suitable. Other molecular classes such as peptides may be most successfully analysed

with acetonitrile and higher aqueous percentage, the addition of organic acids or bases have

also been used to aid in the ionisation process [1].

During the analysis the surface is wetted by the spray, forming a thin liquid film into which

species are solubilised and then ejected from the surface as secondary droplets, due to the

continued gas flow. These secondary droplets will contain molecules from the surface, many

of which will be ionised due to the charge imparted on the primary droplets. An inlet capillary

which is attached to the first vacuum region of the mass spectrometer draws the droplets

from above the surface, and through evaporation of the solvent will present molecular ions to

the mass analyser.

By moving the sample under this spray and recording the co-ordinates, chemical maps of

the surface can be reconstructed that can be retrospectively interrogated. Using the time-of-

flight mass analyser, all molecules that are collected and ionised by the spray will be

detected, there is no need for prior knowledge of the molecules of interest and no labels are

2

required. The trade-off is that the data sizes are large and typically multivariate analysis

techniques are required to draw out the important information.

Conventionally, DESI-MSI is performed by moving the sample horizontally under the spray

to collect a linescan, and then the sample is moved down one y-pixel step size, and the next

linescan is conducted. As the sampling is a continuous event the spray is always on, the size

of the pixels in the x-direction is dependent upon the rate of the mass spectrometer scanning

and the velocity of the sampling stage. If the stage is moving at 100 µm/s and the mass

spectrometer collects 1 scan per second then each pixel in the image will have an x pixel

size of 100 µm. If the stage is moving at 1mm/ s and the mass spectrometer has a scan time

of 0.1s then the x-pixel length will still be 100 µm, and experiment will be conducted 10 times

quicker, though each pixel will be the accumulation of ten times less ion signal. If the species

of interest are abundant then this will not be a problem, however for lower concentration

molecules (or those with a lower ionisation efficiency) then longer experiments may present

more meaningful results. It is typical to set the distance in between each linescan (the length

of the y-pixel) to be the same as the x pixel length, therefore in the examples above the

stage would be moved 100 µm on the y axis each time.

Data viewing and analysis

A single MSI experiment can contain tens or hundreds of thousands of pixels, with each one

containing a whole mass spectrum, and as such it is not uncommon for one image to contain

10 GB or more of data. In order to make this manageable, the data is processed by

combining all spectra and carrying out a peak picking routine. The user chooses how many

peaks to keep for visualising, and these peak areas are extracted to form the final format

which is a matrix of co-ordinates, m/z values and intensities. For the Waters data this can be

viewed with the High Definition Imaging (HDI) 1.4 software.

Alternatively, there are a number of software packages that can accept the standard imzML

file format, to which all data can be converted. In this manuscript the data was converted to

the imzML format and then imported into the SCiLS Lab software (SCiLS, Bremen,

Germany). This package offers a set of multivariate analysis tools to compare regions of

interest drawn by the user (e.g. epidermis vs dermis) or to perform unsupervised

segmentation of the data based on the chemical profile of the pixels. Further tools then allow

the significance of differentiating ion species using hypothesis testing and received operating

characteristic curves. When multiple samples are to be compared, such data handling

approaches are invaluable.

Mass spectrometry analysis

The DESI-MSI experiments were performed on a Xevo-G2-XS Q-ToF instrument (Waters,

Wilmslow, UK) with a Prosolia (Indianapolis, US) 2D DESI source. The analyses were

conducted in both positive and negative ion mode.

The final DESI conditions chosen were 98:2 (v/v) methanol:water solvent flowing at 2µL/min

with a nebulising gas of 6 bar N2. The inlet capillary to sprayer distance was 6mm, 75°

sprayer to surface angle, 5° inlet capillary collection angle and a sprayer to surface distance

1mm. In order to maximise ion intensity yields a heated inlet capillary was constructed and

operated at approximately 450°C powered by a 20V DC power supply. A 2.5mL Hamilton

syringe and a Harvard 11 Elite (Harvard Scientific) syringe driver system were used to

supply the solvent flow, with operating times of approximately 20 h between refilling.

3

Experiments were set up by first collecting an optical snapshot of the slide on the stage

using a Logitech C920 webcam, and registering this in the HDI software. Regions to be

analysed can be drawn on this optical image and used to set up the experiment. Additional

parameters also defined at this stage are the polarity to be analysed, the pixel dimensions

required and the rate of sampling for the experiment.

References

1. Wiseman J M, Ifa D R, Venter A, Cooks R G. Nat Protocols 2008: 3: 517-524.