PNPLA1 defects in patients with autosomal recessive congenital ichthyosis and KO mice sustain PNPLA1 irreplaceable function in epidermal omega-O-acylceramide synthesis and skin permeability barrier

Abstract

Autosomal recessive congenital ichthyosis (ARCI) is a heterogeneous group of monogenic genodermatoses that encompasses non-syndromic disorders of keratinization. The pathophysiology of ARCI has been linked to a disturbance in epidermal lipid metabolism that impaired the stratum corneum function, leading to permeability barrier defects. Functional characterization of some genes involved in ARCI contributed to the identification of molecular actors involved in epidermal lipid synthesis, transport or processing. Recently, PNPLA1 has been identified as a gene causing ARCI. While other members of PNPLA family are key elements in lipid metabolism, the function of PNPLA1 remained unclear. We identified 5 novel PNPLA1 mutations in ARCI patients, mainly localized in the putative active enzymatic domain of PNPLA1. To investigate Pnpla1 biological role, we analysed Pnpla1-deficient mice. KO mice died soon after birth from severe epidermal permeability defects. Pnpla1-deficient skin presented an important impairment in the composition and organization of the epidermal lipids. Quantification of epidermal ceramide species highlighted a blockade in the production of ω-O-acylceramides with a concomitant accumulation of their precursors in the KO. The virtually loss of ω-O-acylceramides in the stratum corneum was linked to a defective lipid coverage of the resistant pericellular shell encapsulating corneocytes, the so-called cornified envelope, and most probably disorganized the extracellular lipid matrix. Finally, these defects in ω-O-acylceramides synthesis and cornified envelope formation were also evidenced in the stratum corneum from PNPLA1-mutated patients. Overall, our data support that PNPLA1/Pnpla1 is a key player in the formation of ω-O-acylceramide, a crucial process for the epidermal permeability barrier function.

Introduction

The outermost layer of the skin, the stratum corneum (SC), is the essential interface between the body and the outside environment. It mainly fulfils the permeability barrier function of the epidermis by attenuating the transepidermal loss of water and electrolytes and by preventing entry of toxic or pathogenic agents. The SC results from the stacking of corneocytes, the flattened dead cells resulting from epidermal terminal differentiation of the underlying keratinocytes. Corneocytes are encapsulated in the cornified envelope (CE), a highly insoluble protein shell covalently linked on its extracellular side to a lipid monolayer (1,2). These cells are tightly linked to each other by corneodesmosomes and are embedded in a mortar-like lipid extracellular matrix, a complex mixture of Ceramides (Cer), cholesterol and free fatty acids, highly organized in a multilayered lipid structure called lipid lamellae (3,4). Most components of this lipid-rich extracellular matrix are produced by granular keratinocytes. Their precursors such as glucosyl(acyl)Cer, phospholipids and sphingomyelin are stored in the tubulo-vesicular secretory organelles called lamellar bodies. At the stratum granulosum/SC interface, these lipid precursors are released and processed into mature products that form continuous lamellar lipid structures surrounding the corneocytes (5).

Cer of the SC have a complex composition (Supplementary Material, Table S1) (6) with a high level of ω-OH-Cer with an ultra-long acyl chain (C28-C36) (7), which is to a great extent ω-esterified with fatty acids, predominantly linoleic acid, to give rise to ω-O-acylCer (8). These latter epidermal-specific Cer is essential for lipid-matrix organization into lamellae and for the formation of the corneocyte lipid envelope since it is the precursor of protein-bound Cer (2,7). Although the SC lipids play a major role in the skin’s vital properties, some mechanisms of lipid production are still to be described. The recent identification of genes involved in inherited lipid metabolism disorders provided new insights into pathways leading to the synthesis, transport, maturation and/or organisation of SC lipids (9,10).

There is growing evidence that the pathophysiology of Autosomal Recessive Congenital Ichthyoses (ARCI) is closely linked to a disturbance in SC lipid metabolism. ARCI are rare non-syndromic ichthyoses belonging to the heterogeneous group of Mendelian disorders of cornification (11). At birth many patients are “collodion babies”. The skin phenotype subsequently consists in generalised scaling and variable erythroderma, with a wide spectrum of clinical presentations from lamellar ichthyosis to congenital ichthyosiform erythroderma. To date, mutations associated with ARCI have been described in 10 genes: ABCA12 (ARCI4; MIM 601277/242500), ALOX12B (ARCI2; MIM 242100), ALOXE3 (ARCI3; MIM 606545), CERS3 (ARCI9; MIM 615023), CYP4F22 (ARCI5; MIM 604777), LIPN (ARCI8; MIM 613943), NIPAL4 (ARCI6; MIM 612281), PNPLA1 (ARCI10; MIM 615024), SDR9C7 and TGM1 (ARCI1; MIM 242300) (12,13). Some of them have been addressed in many studies which have demonstrated their involvement in SC lipid metabolism. However, the function of the proteins encoded by NIPAL4, LIPN, SDR9C7 and PNPLA1 remains unclear.

Patatin-like phospholipase domain containing 1 (PNPLA1) is one of the 9 members of the PNPLA family, characterized by a highly conserved “patatin” domain. These proteins have diverse lipolytic and acyltransferase activities and play a key role in lipid metabolism (14,15). PNPLA1 is the less characterized member of this family. In human, the protein is expressed in the epidermis, predominantly in the granular layer (16,17). In mice, it was very recently reported that absence of functional Pnpla1 impaired the generation of ω-O-acylCer leading to a lack of functional corneocyte-bound lipid envelope (18). Impairment of ω-O-acylCer synthesis was confirmed using a PNPLA1-deficient cultured human keratinocytes.

Here, using mice with targeted inactivation of Pnpla1, we confirm that the absence of Pnpla1 led to neonatal lethality due to severe epidermal permeability defects. This was accordingly associated to a blockade of ω-O-acylCer synthesis that led to a profound impairment of the cornified lipid envelope. Electron microscopic observations, further suggested an impairment of the intercorneocyte lipid organization. In addition, our detailed profiling of epidermal sphingolipids revealed an increase in 1-O-acylCer production in Pnpla1-deficient epidermis. Finally, we describe five novel PNPLA1 mutations in patients with ARCI. Using SC from PNPLA1-mutated patients, we directly demonstrated a blockade of ω-O-acylCer synthesis and an impairment of CE lipid coverage.

Results

Identification of novel PNPLA1 mutations in patients suffering from ARCI

A total of 5 patients from 3 non-consanguineous Caucasian families exhibited common characteristic features of ARCI (Table 1). There was no significant intra familial variability. Patients from family I were born as collodion babies. The disease remained stable over time for the majority of patients, whereas some reported a mild improvement. The patients were aged between 7 and 65 years. Some of them were treated with acitretin during certain periods of their life. All had a similar phenotype with diffuse ichthyosis of moderate severity, comprising fine whitish or soft brown scales with mild erythroderma and no keratoderma of the palms and soles. Some patients also presented with mildly brown hyperkeratotic skin in the folds (Fig. 1A–F). Remarkably, the affected siblings from family II suffered from keratoconus and also had ectropion. Histopathological examination of skin biopsies from affected individuals showed acanthosis, hypergranulosis and compact hyperkeratosis (Fig. 1G–H).

Figure 1

Clinical and histological features of patients with PNPLA1 mutations. (A,B) Ankles and feet skin in patient I.2. (C,D) Neck skin in patient III.1. (E,F) Skin of the upper body in patient II.2. (B,D,F: enlargement of corresponding surrounded area in A,C,E). (G,H) Hematoxylin–eosin stained sections from healthy donor skin (control) and PNPLA1-mutated patient III.1 (scale bars = 50 µm).

Table 1

Clinical and genetic data of patients suffering from ARCI

Family	Patient	Sex (M/F)	Age at evaluation (Y)	Collodion baby at birth	Improvement over time	Ectr- opion	Hyperk- eratosis	Acitretin in the past	PNPLA1 Nucleotide varianta	PNPLA1 Amino acid changeb
I	1	M	12	+	+	–	+	–	c.[418T>C];[=], c.[820_820delC];[=]	[p.Ser140Pro];[=], [p.Arg274Glyfs*7];[=]
	2	F	7	+	+	–	+	–	c.[418T>C];[=], c.[820_820delC];[=]	[p.Ser140Pro];[=], [p.Arg274Glyfs*7];[=]
II	1	M	65	–	–	− *	–	+	c.[266C>T];[=], c.[418T>C];[=]	[p.Pro89Leu];[=], [p.Ser140Pro];[=]
	2	F	54	–	–	+ *	–	+	c.[266C>T];[=], c.[418T>C];[=]	[p.Pro89Leu];[=], [p.Ser140Pro];[=]
III	1	F	12	–	–	–	+	–	c.[335C>A];[=], c.[350C>T];[=]	[p.Ser112Tyr];[=], [p.Thr117Met];[=]

^aReference sequence PNPLA1, NM_001145717.1.

^bReference sequence PNPLA1: NP_001139189.2.

F, female, M; male; Y, years; *, keratoconus.

These clinical and histological features indicated ARCI. Mutation screening of the patients’ genomic DNA was performed by next generation sequencing for patient I-1, II-2 and III-1. No disease-causing variations were detected in the genes TGM1, NIPAL4, ALOX12B, CYP4F22, ALOXE3, ABCA12 and LIPN, triggering ARCI and present in our custom panel. However, the 3 patients had two heterozygous mutations in PNPLA1 (Supplementary Material, Table S2). These mutations were confirmed by Sanger sequencing of the corresponding exons in the patients, as well as in the siblings of patient I-1 and patient II-2 (Table 1). We confirmed that the mutations were compound heterozygous by sequencing the corresponding exons from their respective parents or children. This was consistent with the inheritance mode from non-consanguineous parents. These five novel mutations as well as the previously reported PNPLA1 mutations are detailed in Table 2 and Supplementary Material, Fig. S1 (17,19–21). Three of the novel mutations were already referenced in some databases but showed in the Exome Aggregation Consortium (ExAC) a frequency < 1/10,000 with no homozygous individuals for the alternate allele. The deletion c.820_820delC is predicted to lead to a frameshift and to generate a premature stop codon 24 nucleotides downstream of the deletion (p.Arg274Glyfs*7). This mutation can be predicted to result in either mRNA decay or in the synthesis of a truncated protein. The four missense mutations were located in the patatin-like domain of PNPLA1. Their impact on PNPLA1 structure and function was assessed using the in silico protein prediction tools Polyphen-2 (22) and SIFT (23). Both analyses gave concordant results with damaging consequences for all the mutations except variation Pro89Leu for which a deleterious consequence was predicted by Polyphen-2 only. Altogether, these data strongly suggest that these newly identified mutations are involved in the clinical phenotype of the patients.

Table 2

Novel (this study) and known PNPLA1 mutations associated with ARCI

Nucleotide changea	Amino acid changea	Exon	Type of variation	Protein domain	Polyphen-2.1b	SIFTb	dbSNP no.	MAF in ExAC	References
c.56C>T	p.Ser19Leu	Exon 2	Missense	patatin-like	1	0	–	–	(19)
c.100G>C	p.Ala34Pro	Exon 2	Missense	patatin-like	1	0.01	–	–	(19)
c.100G>A	p.Ala34Thr	Exon 2	Missense	patatin-like	0.979	0.01	–	–	(21)
c.176C>T	p.Ala59Val	Exon 2	Missense	patatin-like	1	0.02	–	–	(17)
c.266C>T	p.Pro89Leu	Exon 3	Missense	patatin-like	0.993	0.13	–	–	This report
c.335C>A	p.Ser112Tyr	Exon 3	Missense	patatin-like	0.996	0.01	rs369445146	8.239e-06	This report
c.350C>T	p.Thr117Met	Exon 3	Missense	patatin-like	1	0	rs371307766	2.473e-05	This report
c.374C>A	p.Thr125Asn	Exon 3	Missense	patatin-like	1	0	–	–	(19)
c.387C>A	p.Asp129Glu	Exon 3	Missense	patatin-like	0.983	0.012	–	–	(20)
c.391G>T	p.Glu131*	Exon 3	Nonsense	patatin-like	–	–	–	–	(17)
c.418T>C	p.Ser140Pro	Exon 3	Missense	patatin-like	0.957	0	rs781053760	2.148e-05	This report
c.421A>G	p.Lys141Glu	Exon 3	Missense	patatin-like	0.982	0.02	–	–	(19)
c.488C>T	p.Pro163Leu	Exon 4	Missense	patatin-like	1	0.07	–	8.242e-06	(19)
c.514G>A	p.Asp172Asn	Exon 5	Missense	patatin-like	1	0	rs373148099	8.357e-06	(19)
c.820_820delC	p.Arg274Glyfs*7	Exon 7	Frameshift	central region	–	–	–	–	This report

^aReference sequences PNPLA1: NM_001145717.1, NP_001139189.2.

^bPolyphen-2: http://genetics.bwh.harvard.edu/pph2/; SIFT: http://sift.jcvi.org/www/SIFT_enst_submit.html.

dbSNP, Single Nucleotide Polymorphism Database; MAF, minimum allele frequency; ExAC, Exome Aggregation Consortium.

Pnpla1 invalidation in mice induces a lethal phenotype with major defects in the epidermal barrier

In order to better understand the function of PNPLA1 in the epidermis, we developed Pnpla1 knockout (KO) mice on a C57BL/6 background (Supplementary Material, Fig. S2). KO mice were identified using a PCR-based genotyping strategy and absence of a detectable level of Pnpla1 mRNA was confirmed by RT-PCR analysis (Supplementary Material, Fig. S3A). We checked Pnpla1 expression in various murine tissues by quantitative RT-PCR analysis of wild type (WT) embryos. As expected, the highest Pnpla1 mRNA level was detected in the skin. About half was expressed in the stomach with very low or undetectable levels in the other tissues we analysed (Supplementary Material, Fig. S3B). Thus, in mice, Pnpla1 seems to be mainly expressed in cornified squamous epithelia. As Pnpla1 KO mice were obtained by insertion of a promoterless cassette including LacZ in the first intron of the endogenous Pnpla1 gene, the activity of the endogenous Pnpla1 promoter could be analysed in KO mice by X gal coloration assays. LacZ-reporter gene expression was thereby detected in the skin and cornified stratified squamous epithelium of the digestive tract from KO mice, thus confirming the results obtained by quantitative RT-PCR (Supplementary Material, Fig. S3C). In particular, a blue labelling in the epidermis was observed from the granular layer, consistent with the late expression of Pnpla1 during keratinocyte terminal differentiation previously reported in humans and dogs (16,17). Persistence of the staining in the SC most probably related to β-galactosidase stability, since no longer transcription nor translation occurred in this epidermal layer.

Heterozygous Pnpla1^+/tm1a mice were phenotypically indistinguishable from WT mice and reproduced normally. Pnpla1^tm1a mice were born at the expected Mendelian ratio (WT 28%, heterozygous 47% and homozygous 25%, n = 213), but died soon after birth. They were easily recognized by the appearance of their skin, which was taut and shiny without normal skin folds. They showed a reduced mobility and the milk stripe was absent. E18.5 embryos obtained by caesarean delivery showed the same phenotype. We observed eversions of the lips (eclabium) (Fig. 2A), and the weight of the KO mice was significantly reduced compared to WT and heterozygous embryos, although this difference was minor (data not shown). Histological examination of the skin revealed mild acanthosis (Fig. 2B, upper panel). This was consistent with faint staining of the suprabasal keratinocytes from KO mice with anti-keratin 6 antibody, whereas the immunolocalisation of the cell proliferation marker Ki67 in the basal layer of the WT and KO epidermis was similar (Fig. 2B, lower panel). The most striking histological difference in comparison with the WT epidermis was a thick, compact SC without the normal basket-weave appearance. The keratohyalin granules also appeared smaller and less numerous in the Pnpla1^tm1a than in the WT skin (Fig. 2B, upper panel). In agreement, quantitative RT-PCR analysis, western blot and immunochemical analysis revealed a decrease in the expression of the late differentiation proteins filaggrin and loricrin, normally stored in these granules (Fig. 2C, Supplementary Material, Fig. S4). In contrast, we did not observe any difference in the mRNA level of involucrin, keratin 10, corneodesmosin and desmoglein 1 between WT and KO skin (Supplementary Material, Fig. S4A). However, higher corneodesmosin levels were detected by western blot in the KO skin, most probably resulting from the retention of corneodesmosomes in the compact and hyperkeratotic SC (Fig. 2C).

Figure 2

Pnpla1 invalidation in mice induces a lethal phenotype with major defects in the epidermal barrier (A) Skin and lips appearance of WT and Pnpla1^tm1aE18.5 embryos obtained by caesarean delivery. (B) Hematoxylin–eosin (H&E) coloration of skin sections from WT and Pnpla1^tm1aE18.5 embryos (scale bars = 100 µm and 30 µm for low and high magnification images, respectively) (upper panel) and immunohistochemical staining of skin sections from WT and Pnpla1^tm1a E18.5 embryos using antibodies specific to Ki67 and keratin 6 (K6) (n = 3 for both genotype) (scale bars = 100 µm) (lower panel). (C) Western blot analysis of Filagrin (FLG), Loricrin, Involucrin, Corneodesmosin (Cdsn) and Actin in protein extracts from WT and Pnpla1^tm1a E18.5 embryos epidermis. (D) Barrier-dependent dye exclusion assay (toluidine blue) performed on WT and Pnpla1^tm1aE18.5 embryos (n = 8 for WT and n = 7 for Pnpla1^tm1a). (E) Transepidermal Water Loss (TEWL) assay performed on whole body of E18.5 embryos (n = 30 for WT and n = 25 for Pnpla1^tm1a). Data are presented as mean ± standard deviation. ****P ≤ 0.0001; Student’s t-test.

We assessed the outside-in permeability barrier by performing the dye penetration assay at E18.5 (24). The WT skin was completely impermeable, in contrast to the Pnpla1^tm1a skin which was permeable to toluidine blue dye, as indicated by unequivocal staining (Fig. 2D). We also assessed the inside-out water barrier by measuring the transepidermal water loss (TEWL) on whole embryos. Pnpla1^tm1a embryos showed a significantly increased TEWL in comparison to the WT (greater than two fold) (Fig. 2E). These findings clearly indicate that both the outside-in and the inside-out water barrier function were severely affected in the epidermis of Pnpla1^tm1a mice.

These macroscopic, histological and functional alterations observed in Pnpla1^tm1a murine skin are features that are reminiscent of ARCI and indicate that mice with defective Pnpla1 develop a skin condition similar to human ARCI caused by PNPLA1 mutations.

Absence of Pnpla1 in mice affects keratinocyte lipid organization in the SC

Electron microscopy (Fig. 3) confirmed hyperkeratosis of the skin in Pnpla1-deficient mice, as indicated by the presence of very compact and more numerous horny layers (Fig. 3A and B). Ultrastructural examination of the KO skin also confirmed that both the size and number of keratohyalin granules were reduced in comparison to WT skin (Fig. 3C and D). No difference in the morphology of the corneodesmosomes could be detected, but these junctional structures persisted to the upper layers of the cornified layer of the KO epidermis, consistent with the delay of the desquamation process (Fig. 3E–G). Of note, lipid droplets inside corneocytes were observed along the entire height of the SC (Fig. 3B,H). In addition, deposits resembling extruded contents of lamellar bodies in the transient zone were seen in the intercellular spaces of higher SC levels (Fig. 3I). Besides, the multilayer lipid structure, the lamellae, with the typical alternating electron-dense/electron-lucent repeat pattern were visualised in the SC of WT mice after ruthenium tetroxide post-fixation (n = 5) in an alternative manner, depending on the orientation of the section with respect to the electron beam. In contrast, we never observed such highly organized lamellae in the SC of Pnpla1^tm1a mice (n = 3) but only diffuse, loosely organized lipid structures in the intercellular spaces (Fig. 3J-L). We analysed lipid distribution in the epidermis by Oil Red O staining (Fig. 4A). We visualised some spots in the intercorneocyte space of the Pnpla1-deficient epidermis which were not observed in the WT epidermis. This uncommon accumulation of lipids in pearl-like structures was consistent with the electron microscopy observations in Pnpla1-deficient mice which suggested a disorganisation of the intercorneocyte lipids in the SC. Finally, to assess the impact of Pnpla1 deficiency on covalently bound SC lipids, we purified CEs from WT and mutant epidermis and evaluated their maturity by a combination of Nile Red staining and involucrin immunostaining (Fig. 4B). The CEs from WT mice were strongly and homogeneously stained with Nile Red, but were virtually all involucrin-negative. Conversely, only a few CEs from KO epidermis were stained with Nile Red whereas most of them strongly reacted with the anti-involucrin antibody. This indicated that WT mice had mainly mature hydrophobic CEs with a covalently linked external lipid monolayer, while CEs from KO mice presented defective lipid coverage and were essentially composed of crosslinked proteins.

Figure 3

Pnpla1 deficiency in mice leads to important ultrastrutural defects. (A–L) Transmission electron microscopy analysis of ultrathin sections of skin pieces (A-I) or cryosections (J–L) from WT (A, C, E, H,J) and Pnpla1^tm1a (B,D,F–I,K.L) E18.5 embryos. (A,B) SC appearance (scale bars = 2 µm); (C,D) Keratohyalin granules inside a granular keratinocyte (arrows) (scale bars = 100 nm); (E–G) Corneodesmosomes in the intercellular space of the SC (scale bars = 50 nm); (B,H) Lipid droplets inside corneocytes (arrows) (scale bars = 50 nm in H); (I) Deposits resembling extruded contents of lamellar bodies in intercellular space of the SC (scale bars = 100nm); (J–L) Lamellae visualization after ruthenium tetroxide post-fixation in the intercellular space of the SC (scale bars = 50 nm). (A-I): WT (n = 6) and Pnpla1^tm1a (n = 6) embryos were proceeded; (J–L): WT (n = 5) and Pnpla1^tm1a (n = 3) embryos were proceeded. Representative images are shown.

Figure 4

Alteration of SC lipids in Pnpla1^tm1a mice and PNPLA1-mutated patients. (A) Oil Red O staining of lipids on skin sections from WT and Pnpla1^tm1a E18.5 embryos (arrows show neutral lipids accumulation; scale bars = 25 µm). (B) Double staining of CEs from the epidermis of WT and Pnpla1^tm1a E18.5 embryos using Nile Red (red) and an antibody specific to involucrin (green) (scale bars = 100 µm). (C) Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) analysis of lipid extract from epidermis of WT and Pnpla1^tm1a E18.5 embryos. Only sphingolipids with the major sphingoid base C18-sphingosine were analyzed. Absolute molar data were normalized to the total amount of all sphingolipids, which were determined. All sphingolipid species quantified in LC-MS/MS are listed in Supplementary Material, Table S8A and B. For other normalization and for zoom in of minor subgroups see Supplementary Material, Figure S5B. (D) Gas-liquid chromatography (GC) analysis of lipid extract from epidermis of WT and Pnpla1^tm1a E18.5 embryos. (A,B and D: n = 3 for both genotypes; C: n = 9 WT, 7 Pnpla1^tm1a). Data are presented as mean ± standard deviation. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; Student’s t-test.

Taken together, these observations showed an obvious impairment of lipid processing and organization in the SC of Pnpla1^tm1a mice.

Pnpla1 is required for ω-O-acylCer synthesis in mice

Since Cer are shared components of the lamellae and the CEs, we assessed the epidermal Cer composition using thin layer chromatography (Supplementary Material, Fig. S5A). Pnpla1-deficient mice had a modified sphingolipid profile with, in particular, disappearance of ω-O-acylCer. Omega-O-acyl fatty acids (FA[EO]), the assumed degradation products of ω-O-acylCer, were also dramatically reduced. LC-MS/MS analysis further showed that in the Pnpla1-deficient epidermis, linoleic acid-esterified Cer (Cer[EOS]) had basically disappeared. Instead, mutant epidermis accumulated ω-OH-Cer (Cer[OS]), an intermediate Cer species detected at low level in WT mice because it is normally processed rapidly in ω-O-acylCer. Parallel to ω-OH-Cer, corresponding glucosylCer (GlcCer[OS]) and corresponding free ω-OH-fatty acid (FA[O]) also accumulated in the mutant mouse epidermis. At the same time, ω-OH-Cer-corresponding sphingomyelins (SM[OS]), in tiny amounts in WT mice, increased in KO mice (Fig. 4C, Supplementary Material, Fig. S5B and C). Moreover, in accordance with reduced ω-O-acylCer levels, we observed a significant albeit moderate increase in free linoleic acid (Fig. 4D). Consistent with the thin layer chromatography analysis, LC-MS/MS revealed a strong decrease in ω-O-acyl fatty acids (FA[EO]) (Supplementary Material, Fig. S5C). Finally, the level of protein-bound Cer (Cer[POS]) and their catabolites, the protein-bound fatty acids (FA[PO]), was drastically reduced (Fig. 4C and Supplementary Material, Fig. S5C). As both species essentially derive from the lipid CE, this confirmed the defective lipid coverage of the CEs from Pnpla1-deficient mice observed by Nile Red/involucrin double staining.

Furthermore, although only a tendency, mutant mice doubled the proportion of 1-O-acylCer, a group of Cer esterified to very long acyl chains in the 1-O- position (25). In contrast to normal skin, ω-OH-Cer was now also found to be 1-O-esterified (Cer[1-O-EOS]).

The perturbations in Cer composition caused by Pnpla1 deficiency were not correlated with significant changes in the expression level of genes involved in epidermal lipid metabolism, as shown by qRT-PCR analysis (Supplementary Material, Fig. S5D).

Lipid perturbations revealed in mice are also present in PNPLA1-mutated patients with ARCI

In order to examine whether specific changes in SC lipids evidenced in the Pnpla1^tm1a mice were also present in our patients suffering from ARCI, we collected non-invasively successive tape strips from the skin of patients II.1, II.2 and healthy donors. CEs prepared from these strips were assessed for their maturity by Nile Red/involucrin double staining (Fig. 5A). CEs from healthy donors were mainly stained with Nile Red while a few were involucrin-positive. In contrast, almost all CEs prepared from patient II-1 reacted with the anti-involucrin antibody. Cer composition in the SC of healthy individuals and patients was examined by LC-MS/MS analysis of lipids prepared from the strips (Fig. 5B). As observed in Pnpla1-deficient mice, the results showed a dramatic reduction in ω-O-acylCer and an accumulation in ω-OH-Cer in the SC of patients II-1 and II-2. These latter observations, together with the strong resemblance of the skin phenotype between PNPLA1-mutated patients and Pnpla1^tm1a mice, corroborate an essential role of PNPLA1/Pnpla1 in epidermal barrier permeability through its involvement in ω-O-acylCer synthesis.

Figure 5

Lipid perturbations evidenced in mice are also present in PNPLA1‐mutated patients with ARCI. (A) Double staining of CEs from healthy donors (control) and PNPLA1-mutated patient II.1 using Nile Red (red) and an antibody specific to involucrin (green) (scale bars = 100 μm). (B) Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) analysis of lipid extract from tape stripes from healthy donors or PNPLA1-mutated patients with ARCI. Only Sphingolipids with a C18-sphingosine were recorded. Fold changes of Cer species were determined by using an internal standard (see Methods). (A: n =  3 healthy donors each 1 sample, patient II.1, 3 samples; (B) n =  6 healthy donors each 1 sample, patients II.1 and II.2 each 3 samples). Data are presented as mean ± standard deviation. ***P ≤ 0.001 Student’s t-test.

Discussion

Pnpla1 deficiency in mice led to neonatal lethality. Pnpla1^tm1a E18.5 embryos and newborns had a thick, taut and shiny skin with a shellacked appearance. This “collodion-like” appearance was strongly evocative of the collodion baby observed in humans. This phenotype was associated with impairment in the outside-in and inside-out permeability barrier function. Furthermore, the tightness of the skin led to reduced mobility and thus failure to suckle maternal milk. Death usually occurs in a window of 12-24 h after birth in the case of non-feeding newborn mice (26). Pnpla1-deficient mice died mostly within the first 12 h of life. Thus, Pnpla1-deficient mouse lethality was most probably due to severe dehydration caused by both an inability to feed and epidermal barrier impairment. Similar lethal phenotypes have also been reported in mice invalidated for other genes causing ARCI (27–31). Concerning patients suffering from ARCI, a survey reported that mortality was as high as 50% in 1960 but nowadays is only 5% (32). Thus, the discrepancy in phenotype severity between humans and mice could result from the fact that, in contrast to mice, patients born as collodion babies benefit from intensive care.

ω-O-acylCer are an epidermal-specific Cer species that are essential for the formation and maintenance of the epidermal barrier. Their production initially requires ultra-long chain (C26 to C36) fatty acid synthesis. This first step depends on fatty acid elongases encoded by ELOVL1 and ELOVL4 (12,33–35). Then CYP4F22, a member of the P450 cytochrome superfamily, preferentially ω-hydroxylates C28-C36 ultra-long chain fatty acids and the latter are subsequently used by Ceramide synthase-3 to produce ω-OH-Cer (28,36). Finally, a hitherto unknown acyltransferase catalyses the formation of an ester bond between fatty acid, predominantly linoleic acid, and the ω-OH-Cer to form ω-O-acylCer.

In this report, we confirm that Pnpla1-deficient mice had a blockade in ω-O-acylCer synthesis. Quantification of different species of Cer in the mutant epidermis highlighted a drastic reduction in ω-O-acylCer with a concomitant accumulation of ω-OH-Cer. A similar lipid profile was described in mice with triglyceride lipase cofactor Abhd5 deficiency (37,38). Abhd5 and Pnpla1 are both required for ω-O-acylCer synthesis from ω-OH-Cer. In contrast to adipocyte triglyceride metabolism where it acts as a cofactor of adipocyte triglyceride lipase (ATGL/Pnpla2), Abhd5 appears not to act likewise in epidermal triglyceride metabolism (38), leaving its molecular role in ω-O-acylCer metabolism open. Regarding PNPLA1, it has been shown that it is not a triglyceride hydrolase (17,18). In accordance, our Pnpla1-deficient mice have a slight but significant increase in free fatty acid suggesting a defect in the incorporation rather than in the release of fatty acid. In conclusion, we propose that Pnpla1 could be involved in free fatty acid incorporation on ω-OH-Cer, the last step in ω-O-acylCer production. However, an accurate characterization of PNPLA1 activity at the molecular level is required to confirm this hypothesis.

As recently shown by Grond et al. (18), impairment of ω-O-acylCer was associated with an impairment of cornified lipid envelope in our Pnpla1-deficient mouse model. Our electron microscopy observations further suggest that those lipid abnormalities led to a defective intercorneocyte lipid organisation with absence of the typical lamellae structure. These observations are consistent with the important role of ω-O-acylCer in the arrangement of intercorneocyte lamellae already reported in literature (39,40).

We also report that the lack of virtually all ω-O-acylCer was associated with a slight increase in 1-O-acylCer in the Pnpla1-deficient epidermis. Interestingly, this includes the de novo appearance of 1-O-acylCer containing an ω-OH-Cer backbone (Cer[1-O-EOS]) derived from ω-OH-Cer accumulation. However, this latter represents less than 5% of all 1-O-acylCer. 1-O-acylCer is a new class of epidermal Cer esterified to (very) long acyl chains in 1-O- position, recently identified in humans and mice (25). 1-O-acylCer increase has previously been observed in barrier-deficient GlcCer-synthase deficient (keratinocyte-specific) mice (25) as well as in CerS3-deficient mice (unpublished data). Due to its predictive physicochemical properties, 1-O-acylCer has been proposed to contribute to water permeability barrier function. However, in our model, the overall increase in 1-O-acylCer is unable to compensate the drastic reduction of ω-O-acylCer. Nevertheless, it would be interesting to further investigate the role of 1-O-acylCer in the SC.

We identified 5 novel PNPLA1 mutations in patients with ARCI. The patatin-like domain of PNPLA1 possesses the characteristic features shared by all members of the PNPLA family (14). Notably, it contains an active site with a serine-aspartate catalytic dyad and an oxyanion hole that stabilizes the enzyme-substrate transition state (41) (Supplementary Material, Fig. S1). It also presents a conserved core module where the nucleophilic serine is located in a tight turn between a β-sheet and an α-helix in a well conserved β-β-α-β core structure. The 4 novel missense mutations we identified occur in residues that are located in the patatin-like domain of PNPLA1, like the previously described mutations. None of these substitutions directly affected the catalytic dyad but conserved positions, which were predicted to be deleterious for the encoded protein. The fifth mutation we identified was a frameshift leading to a predicted protein deleted of the C-terminal part of the full-length PNPLA1. This latter region contains a proline-rich domain which is conserved among the adiponutrin-like subgroup of the PNPLA family (PNPLA1-PNPLA5) and involved in lipid binding. In particular, some PNPLA2 mutations causing Neutral Lipid Storage Disease (NLSD) lead to a loss of the C-terminal domain of the encoded protein. This prevented the proper localization of the enzyme to lipid droplets (42,43). Furthermore, a homozygous nonsense mutation that led to the loss of 74 amino acids in the highly conserved C-terminal region of PNPLA1 caused ARCI in dogs (17). Thus the frameshift mutation found in patients from family I most probably leads to a non-functional protein. Finally, our data suggest that PNPLA1 mutations impaired PNPLA1 activity. Indeed using SC from PNPLA1-mutated patients, we directly demonstrate a blockade of ω-O-acylCer. Interestingly, we revealed a subsequent impairment of the lipid coverage of CE purified from PNPLA1-mutated patients.

In conclusion, detailed analysis of Pnpla1-deficient mice and analysis of SC from PNPLA1-mutated patients allow us to validate that PNPLA1/Pnpla1 is an essential actor necessary for the formation of ω-O-acylCer, an epidermis-specific Cer that plays a crucial role in the establishment and maintenance of the skin permeability barrier. This insight into the metabolism of SC lipids in the normal epidermis enhances our understanding of the pathophysiology of ARCI.

Materials and Methods

Subjects and primary samples

Skin biopsies and blood samples were collected for diagnosis and gathered in a biological collection (n°DC-2011-1388, French National Ethics Committees). Clinical and genetic information on the subjects were reported in Table 1. Written informed consent was obtained from all patients or from their legal representative in the case of minors, allowing us to use the taken biological specimens collected for research purposes. Consequently, the work presented in this article did not have to be submitted to the local Ethics Committee.

Molecular genetic analysis

DNA was isolated by standard procedures from peripheral white blood cells of the patients and their parents. Mutation screening was performed by next generation sequencing using the Personal Genome Machine (PGM, Ion Torrent, Thermo Fisher Scientific), and the AmpliSeq technology (Ion Torrent). Using the AmpliSeq Designer tool (ampliseq.com; v2.0), an AmpliSeq Custom Panel was designed to cover 13 genes known to be involved in ichthyoses, comprising exonic regions, exon-intron boundaries as well as 3’ and 5’ Untranslated Regions (UTRs), i.e. a target region of 89.25 kb (Supplementary Material, Table S3). A total of 590 primer pairs, covering 92.14% of the target, were designed and synthesized.

Library construction and sequencing were performed at the GeT-Purpan core facility (Genome and Transcriptome, GenoToul, France). In brief, genomic DNA samples were checked for purity and quantity using Nanodrop (Thermo Fisher Scientific) and a concentration was estimated using Qubit v3 fluorometer (Thermo Fisher Scientific) and Qubit dsDNA HS Assay Kit (Invitrogen). Libraries were produced using 10 ng of genomic DNA for each sample and each pool of AmpliSeq primers, with the Ion AmpliSeq library kit v2.0 (Ion Torrent), and Ion Xpress Barcodes, following the guidelines of the supplier. Final libraries were individually controlled and quantified on High Sensitivity DNA chips of BioAnalyzer (Agilent Technologies). Libraries from a total of 16 patients were then pooled, templated on Ion Sphere Particles using One Touch 2 instrument with HiQ kit, and finally sequenced on the Ion Torrent PGM using the sequencing HiQ chemistry, 200 bases run workflow, on a 316 v2 Ion chip.

The generated data produced were processed using Ion Torrent Suite Software (Thermo Fisher Scientific) for quality filtering, trimming, demultiplexing, aligning on the AmpliSeq design reference, and variant calling. 285,632 usable reads were produced for the 3 patients and the coverage per base was >30X for more than 95% of the bases. Variant caller files were then transferred using Ion Reporter Software (v5.0, http://ionreporter.lifetechnologies.com/, Thermo Fisher Scientific) for filtering and annotation of detected variants.

The PNPLA1 variants were confirmed by Sanger sequencing. In brief, the corresponding exons and intron-exon boundaries were amplified by polymerase chain reaction using specific primers (Supplementary Material, Table S4) and sequenced using the BigDye Terminator v3.1 chemistry (Applied Biosystems) and the ABI3130XL Genetic Analyzer (Applied Biosystems) available on the GeT-Purpan Core facility (GenoToul, France).

All variants have been submitted to NCBI ClinVar Database (https://www.ncbi.nlm.nih.gov/clinvar).

Generation of Pnpla1-deficient mice

Pnpla1-deficient (B6NTac;B6N-A^tm1BrdPnpla1^tm1a(K°^MP)Wtsi/Ics, abbreviated to Pnpla1^tm1a in this report) mice carry a knockout-first allele, in which a promoterless cassette including LacZ and neo genes was inserted in the first intron of the Pnpla1 gene flanked by FRT sites. LoxP sites flank the critical exons (exon2 and 3 of Pnpla1 gene in knockout-first design) (Supplementary Material, Fig. S2). See http://www.mousephenotype.org/data/search/gene?kw=pnpla1 for more details.

Animals

All experiments with animals were approved by local ethic committee UMS006 CEEA-122 and carried out according to our Institutions Guidelines and EU legislation. Mice were housed and maintained in the animal facilities (UMS006 Inserm, Toulouse) under pathogen-free conditions. Mice were killed by Cervical dislocation and decapitation. All efforts were made to minimize suffering.

Timed-matings were performed. Female mice in matings were checked for vaginal plugs and the gestational stage was estimated from the time of a positive plug + 0.5 days. Foetuses were collected on embryonic day 18.5 (E18.5).

Mouse genotyping

Genomic DNA was isolated from or mouse embryo-tails using Nucleospin (Macherey-Nagel). Genotypes were confirmed by PCR using Econotaq Plus Green (Euromedex). The Pnpla1 forward primer (5′- GGGGCACCTTAAGGTGAGGATCGTT -3′) and the Pnpla1 reverse primer (5′- GAAAG CCTGTGACC ACTGAC CC AG -3′) were used and produce a wild type PCR product of 247 bp and a cKO allele product of 297 bp.

RNA isolation and quantitative RT-PCR

Total RNAs from murine tissues were extracted with RNeasy (Qiagen). The quantity of RNA was determined using the Nanodrop 1000 (Thermo Scientific). Quality was assessed using the Agilent 2100 Bioanalyser System with the RNA 6000 Nano kit. All the samples present a Ring Integrity Number (RIN) above 8. Reverse transcription was carried out using the PrimeScript II 1st strand cDNA Synthesis Kit (Takara). Amplification assays were performed with the 7300 Real Time PCR system using the Sybr qPCR SuperMix with ROX (Invitrogen) and specific primer (Supplementary Material, Table S5). Fluorescence was quantified as Ct (threshold cycle) values. Samples were analysed in triplicate, with differences between the three Ct values lower than 0.3. Relative levels of gene expression between samples were determined using Hprt expression for normalisation. Specificity was assessed by sequencing the quantitative RT-PCR amplicons.

Antibodies

The commercially available antibodies used in the study are listed in Supplementary Material, Table S6. Mouse monoclonal antibody F28-27, raised against human corneodesmosin and which cross reacts with the mouse counterpart, was used as previously described (44). Mouse monoclonal antibody AHF3, recognizing human (pro)filaggrin, was used as previously described (45).

Light microscopy

Human and murine skin tissues were fixed in 4% formalin in PBS pH 7.4 for 24 h and then embedded in paraffin. For histological analysis, paraffin sections of 5 µm were stained with hematoxylin and eosin and mounted in Eukitt mounting medium (Euromedex). For immunohistological analysis, paraffin or cryosections (5 µm) were blocked in 2.5% normal horse serum (Vector Laboratories) for 1 h at room temperature. For some antibodies, epitope retrieval was carried out for 40 min at 95 °C in Target Retrieval Solution pH 6.1 (DAKO) or in 50 mM glycine-HCl pH 3.5. Sections were then incubated with the primary antibody for 1 h at room temperature. Dilutions of antibodies are listed (Supplementary Material, Table S6). Rabbit and mouse antibodies were detected with the ImmPRESS anti-rabbit IgG and the ImmPRESS anti-mouse IgG (peroxidase) kit (Vector Laboratories), respectively. Peroxidase activity was revealed with the ImmPACT DAB (Vector Laboratories). Sections were counterstained with hematoxylin and mounted in Eukitt mounting medium. For immunofluorescence, Alexa Fluor® 555-conjugated goat anti-mouse IgG (Life technologies) was used at 1:1000 and sections were counterstained with DAPI and mounted in Mowiol.

For lipid staining, 5 µm-thick cryosections were fixed in 4% formalin in PBS pH 7.4 for 30 min at room temperature. Following a brief rinse in PBS and 30% isopropanol, sections were stained with freshly prepared Oil Red O (Sigma-Aldrich) working solution for 5 min at room temperature, rinsed in 30% isopropanol and distilled water, and mounted in Mowiol.

For β-Galactosidase detection with the chromogenic substrate X-Gal (X-Gal staining) 5 μm cryosections were stained using a LacZ Tissue staining kit (InvivoGen) overnight at 4 °C. Sections were then counterstained with hematoxylin and mounted in Eukitt mounting medium.

Images were taken using a Nikon Eclipse 80i microscope equipped with a Nikon DXM 1200C digital camera and NIS image analyses software. For immunofluorescence staining, images were acquired using an Apotome ZEISS Inv microscope equipped with Axiocam HRm Rev.3 and Zen 2012 software.

Electron microscopy

The skin was cut in pieces of ≈1mm², fixed in 2% glutaraldehyde in Cacodylate buffer (0.1 M, pH 7.2, EMS, Hatfield, PA) for 24 h at 4 °C and post-fixed with 1% OsO4 in Cacodyate buffer (Cacodylate 0.1 M, OsO4 1%, EMS) for 1 h at 4° For lamellae visualization, 50 µM skin cryosections were fixed in 2% glutaraldehyde in Cacodylate buffer (0.1 M, pH 7.2, EMS, Hatfield, PA) for 24 h at 4 °C and post-fixed with 1% OsO4 in Cacodyate buffer (Cacodylate 0.1 M, OsO4 1%, EMS) for 1 h at 4 °C followed by two post-fixation with 0.2% RuO4, 0.25% K3Fe (Cn)6 in Cacodylate buffer for 1h at 4 °C.

Samples were then dehydrated in a graded acetone series and embedded in Spurr’s resin. After 48 h of polymerization at 60 °C, ultrathin sections (80 nm thick) were mounted on 75 mesh formvar-carbon coated copper grids. Sections were stained with Uranyless (Delta Microscopies) and lead citrate. Grids were examined with a TEM (Jeol JEM-1400, JEOL Inc) at 80 kV. Images were acquired using a digital camera (Gatan Orius, Gatan Inc).

Skin barrier function assays

Epidermal barrier function was assessed as previously described (46). In brief, TransEpidermal Water Loss (TEWL) was measured on whole embryo using a TM 300 tewameter (Courage & Khazaka electronic). Data are expressed in g/m²/h. For the dye penetration assay, E18.5 embryos were dehydrated in an ascending methanol series up to methanol 100%, rehydrated for 3 min in PBS and stained with 0.1% toluidine blue in PBS for 10 min at room temperature. After washing in PBS for 15 min, the embryos were photographed with a digital camera (Sony DSC-W50).

Epidermal protein extraction and western blot

Dermo-epidermal cleavage of E18.5 embryo skin was performed by heat treatment and epidermal protein was sequentially extracted in three different fractions by the Fast Prep system (MP Biomedicals) using successive extraction buffers (EB) (EB 1: Tris 40 mM, EDTA 10 mM, NP40 1%, pH 7.4; EB2: Tris 40 mM, EDTA 10 mM, urea 8M, pH 7.4; EB3: Tris 40 mM, EDTA 10 mM, urea 8M, DTT 50 mM, pH 7.4). Determination of the protein content was assessed by Bradford quantification (BioRad). Equal volumes of protein extracts were separated by 10% SDS-PAGE and transferred to nitrocellulose membrane. The blots were probed with commercially available primary antibodies diluted as recommended by the manufacturer (Supplementary Material, Table S6) then with horseradish peroxidase (HRP)-conjugated secondary antibody (Life Technologies) at 1:10000. Protein quantities were normalised by actin detection. Detection was performed with ECL-Prime Reagent (GE Healthcare).

Isolation and analysis of cornified envelopes

Embryo mouse skin (1 mm²) or tape stripping performed in patients and healthy controls were boiled in CE isolation buffer (100 mM Tris-HCl pH 8.5, 20 mM DTT, 5 mM EDTA) containing 2% SDS for 10 min at 95 °C, stirring vigourously. CEs were centrifuged at 12,000 g and resuspended in CE isolation buffer containing 2% SDS. This extraction procedure was repeated three times. Purified CEs were then washed three times in the isolation buffer containing 0.2% SDS and stored at 4 °C.

CE maturity was assessed as described (47). In brief, appropriate concentrations of CE suspension were dropped onto a slide-glass and air-dried. They were fixed in acetone at -20 °C for 10 min, and hydrated in phosphate-buffered saline. The primary antibody allowed a reaction for 1 h at room temperature. A secondary antibody was then applied for 1 h at room temperature. After washing, the CEs were stained with Nile red solution (1 µg/ml) for 30 min at room temperature. Slides were then mounted in Mowiol. Images were taken by fluorescence microscopy using a Nikon eclipse 80i microscope equipped with a Nikon DXM 1200C digital camera and NIS image analysis software.

Lipid extraction and liquid chromatography tandem-mass spectrometric (LC-MS/MS) lipid analysis of sphingolipids

For all experiments, investigators were blinded for sample identity. The epidermis was separated from the dermis and extracted for lipids according to Jennemann et al.(28). In brief, the skin was incubated with thermolysin (2h, 37 °C) at pH 7.4. Then the epidermis was isolated with tweezers, cut in small pieces and freeze-dried. Dried tissue was extracted three times with mixtures of chloroform/methanol/water. Afterwards the residual pellet was washed again 3 times with methanol and two times with 95% methanol before protein bound lipids were released with 1 M KOH in 95% methanol at 60 °C within 2 hours and subsequently neutralized with acetic acid. The pooled extracts were desalted with RP-18 cartridges and aliquots corresponding to 0.1 mg epidermis dry weight were mixed with internal lipid standards for analysis by LC-MS/MS using an Aquity I-class UPLC and a Xevo TQ-S “triple-quadrupole” instrument, both from Waters. Lipids were separated on a 100 mm CSH-C18 column (2.1 × 100 mm; 1.7 µm, Waters) using a gradient between 57% solvent A (50% water, 50% methanol) and 95% solvent B (99% isopropanol, 1% methanol), both containing 10 mM ammonium formiate and 0.1% formic acid as additives (Supplementary Material, Table S7). Lipids analysed for Fig. 4C are listed in Supplementary Material, Table S8 and were detected by multi-reaction monitoring (MRM) (Supplementary Material, Table S8). The following internal standards were used for quantification: Cer(d18:1;14:0)*, Cer(d18:1;19:0)*, and Cer(d18:1;31:0)* were used to quantify Cer[NS]-, [AS]-, [OS]- and [POS]-. GlcCer(d18:1;14:0)*, GlcCer(d18:1;19:0)*, GlcCer (d18:1; 25: 0)*,  and GlcCer(d18:1;31:0)* were used to quantify GlcCer [NS]-, [AS]-, and [OS]-. SM (d18:1;12:0) from Avanti Polar lipids, SM(d18:1;17:0) from Avanti Polar lipids, SM(d18:1;10:0-pyrene) from Sigma Aldrich (N-(10-[1-pyrene]decanoyl) sphingomyelin, P-4275) and SM(d18:1;31:0)* were used to quantify SM[NS]-, [AS]-, and [OS]. For these sphingolipids a calibration curve was calculated from the internal standards for each sample in Microsoft-Excel [function “VARIATION”] to consider potential changes of intensities which are due to increasing m/z-ratios of sphingolipid ions. Cer [EOP]- (d18:1;h32:0;18:2) was used to quantify Cer [EOS]-and GlcCer [EOS]-. For quantification a factor of 0.11 was considered, which resulted from the intensity ratio of equal concentrations of internal standard Cer [EOP]-(d18:1;h32:0;18:2) to external standard Cer [EOS]- (d18:1;h32:0;18:2), both of which had been published previously(48). For quantification of GlcCer [EOS]-, the intensity ratio of internal standards Cer [NS]-over GlcCer [NS]-(0.4) was taken in addition into account. Cer [1-O-ENS]- (18:1;d18:1;17:0) from Avanti Polar Lipids was used to quantify Cer [1-O-ENS]-, [1-O-EAS]-, and [1-O-EOS]-as previously published(25). To determine the fold change of Cer [EOS]- in human samples, Cer [1-O-ENS]- (18:1;d18:1;17:0) was used as internal standard. *: Internal standards marked with a * had been synthesized previously (49).

Liquid chromatography tandem‐mass spectrometric (LC‐MS/MS) analysis of ω-esterified and free ω-hydroxy fatty acids

The identical aliquots of the lipid extracts prepared for sphingolipid analysis were also used to determine free ω-hydroxylated fatty acids and ω-esterified fatty acids using the identical LC-MS/MS system and the identical chromatographic conditions as described for sphingolipid analysis. Multiple reaction monitoring was used to selectively detect free ω-hydroxylated fatty acids and ω-esterified fatty acids and the corresponding transition parameters are listed in Supplementary Material, Table S9. MRM screening was performed for the individual compounds listed in Supplementary Material, Table S10. Relative amounts were calculated from the peak area ratios of free (ω-hydroxylated) fatty acids or ω-esterified fatty acids over the internal standard ¹³C₁₈-stearic acid and normalized to the total amount of detected non-hydroxy or alpha-hydroxy fatty acids.

Fatty acid methyl ester (FAME) analysis

The E18.5 embryo epidermis (15–30 mg) was crushed twice in 1.5 ml of methanol/5 mmol/L ethyleneglycol-bis-(β-aminoethylether)-N,N,N′,N′-tetraacetic acid (2:1 vol/vol) using a Fast Prep (MP Biochemicals), then extracted according to Bligh and Dyer (50) in dichloromethane/methanol/water (2.5: 2.5: 2.1, v/v/v), in the presence of the internal standards heptadecanoic acid (2 μg). The lipid extract was methylated in 14% boron trifluoride methanol solution (Sigma-Aldrich, 1 ml) and heptane (1 ml) at RT for 10 min. After adding water (1 ml) to the raw preparation, FAMEs were extracted with heptane (3 ml), evaporated to dryness and dissolved in ethyl acetate (20 μl). FAMEs (1μl) were analysed by gas-liquid chromatography (51) on a Clarus 600 Perkin Elmer system using a Famewax RESTEK fused silica capillary columns (30 m x 0.32 mm i.d, 0.25 μm film thickness). The oven temperature was programmed from 110 °C to 220 °C at an increment of 2 °C per min and the carrier gas was hydrogen (0.5 bar). The injector and the detector were heated 225 °C and 245 °C, respectively.

Thin layer chromatography ceramide profiling

Ceramides from epidermis E18.5 embryos were analysed by means of high-performance thin layer chromatography. The lipid extract were dissolved in chloroform/methanol/water (10:10:1, vol/vol/vol), and deposited on a HPTLC LiChrospher Silica Gel 60 F254S glass plates (Merck), which had been developed with chloroform/methanol (9/1) and dried before. The TLC plates were developed with chloroform/methanol/acetic acid (60/35/8) until a height of 2.5 cm, dried, developed with chloroform/methanol/acetic acid (190/9/1) until a height of 8.5 cm, dried and developed with chloroform/methanol/acetic acid 190/9/1) until the top (10 cm). After drying, the chromatogram was sprayed with 10% (wt/vol) cupric sulphate hydrate in 8% (wt/vol) phosphoric acid and charred by heating at 180 °C for 8 min. Standards are the same as those used for the LC-MS/MS analyses.

Statistical analysis

All analyses were carried out using Microsoft Excel. Variance equality was initially assessed using the F-test (Fisher). Student’s two-tailed t-tests, with or without equal variances, were then used. Data were presented as the mean ± s.d. P-value ≤ 0.05 was considered to be statistically significant.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We wish to thank all of the patients and their families who participated in this study.

We are grateful to Marion Roy, Sabrina Benaouadi and Patrick Aregui for their excellent technical assistance. We also wish to thank the staff of UMS 006, especially Rachel Balouzat, Sylvie Appolinaire and Guillaume Morera from the animal facilities, Florence Capilla and Talal Al Saati from the technical platform of ‘histopathologie expérimentale’. We are grateful to Sophie Allard and Astrid Canivet from the imaging technical platform (Inserm UMR 1043-CNRS UMR 5282, CPTP, Toulouse) and Stephanie Balor from the Multiscale Electron Imaging platform (METi) at the ‘Centre de Biologie Intégrative’. We wish to acknowledge Michel Simon (Inserm UMR1056, UDEAR, Toulouse) for critically reviewing the manuscript.

R.S. was supported by a joint grant (“ZAFH ABIMAS”) from ZO IV by the Landesstiftung Baden-Württemberg and the Europäischer Fonds für regionale Entwicklung (EFRE) to Carsten Hopf, L.O. was supported by the Czech Science Foundation [grant number 16-25687J]. The lipid analyses performed in the Toulouse INSERM Metatoul-Lipidomique Core Facility-MetaboHub were supported by the French National Research Agency [grant number ANR-11-INBS-010].

Conflict of Interest statement. None declared.

Funding

A joint grant (“ZAFH ABIMAS”) from ZO IV by the Landesstiftung Baden-Württemberg, Europäischer Fonds für regionale Entwicklung (EFRE), the Czech Science Foundation [grant number 16-25687J], and French National Research Agency [grant number ANR-11-INBS-010] and the French Society for Dermatology (SFD).