ORIGINAL ARTICLE

Aloxe3 Knockout Mice Reveal a Function of Epidermal -3 as Synthase and Its Pivotal Role in Barrier Formation Peter Krieg1, Sabine Rosenberger1, Silvia de Juanes1, Susanne Latzko1, Jin Hou2, Angela Dick2, Ulrich Kloz3, Frank van der Hoeven3, Ingrid Hausser4, Irene Esposito5, Manfred Rauh2,6 and Holm Schneider2,6

Loss-of-function in the lipoxygenase (LOX) ALOX12B and ALOXE3 are the second most common cause of autosomal recessive congenital . The encoded proteins, 12R-LOX and epidermal LOX-3 (eLOX-3), act in sequence to convert fatty acid substrates via R-hydroperoxides to specific epoxyalcohol derivatives and have been proposed to operate in the same metabolic pathway during epidermal barrier formation. Here, we show that eLOX-3 deficiency in mice results in early postnatal death, associated with similar but somewhat less severe barrier defects and morphological changes than reported earlier for the 12R-LOX- knockout mice. Skin lipid analysis demonstrated that the severity of barrier failure is related to the loss of covalently bound in both 12R-LOX- and eLOX-3-null mice, confirming a proposed functional linkage of the LOX pathway to processing and formation of the corneocyte lipid envelope. Furthermore, analysis of free oxygenated fatty acid metabolites revealed strongly reduced levels of hepoxilin metabolites in eLOX-3-deficient , indicating an additional function of eLOX-3 in mammalian skin as a hepoxilin synthase linked to the 12S-LOX pathway. Journal of Investigative Dermatology (2013) 133, 172–180; doi:10.1038/jid.2012.250; published online 26 July 2012

INTRODUCTION to cause the described phenotype, among them are the Autosomal recessive congenital ichthyosis (ARCI) are a lipoxygenase (LOX) genes ALOX12B and ALOXE3. The cor- clinically and genetically heterogeneous group of severe responding proteins have been proposed to be involved in the keratinization disorders with an estimated prevalence of 1 in formation of the permeability barrier of the skin, an important 100,000 newborns (Oji and Traupe, 2006). Most of the patients protection against water loss and physical injury. are born as collodion babies. Later, the main symptoms are ALOX12B and ALOXE3 encode the 12R- widespread scaling of the skin and varying degrees of erythema. LOX and epidermal LOX-3 (eLOX-3), respectively. Both are To date, mutations in eight different genes have been identified members of the epidermal subfamily of mammalian LOX with preferential expression in the skin and several other epithelial tissues (Krieg et al., 2002). LOXs constitute a 1Genome Modifications and Carcinogenesis, German Cancer Research widespread family of non-heme iron that insert 2 Center, Heidelberg, Germany; Department of Pediatrics, University of molecular oxygen into polyunsaturated fatty acids, producing Erlangen-Nu¨rnberg, Erlangen, Germany; 3Transgenic Service, German Cancer Research Center, Heidelberg, Germany; 4Department of Dermatology, highly active lipid mediators (Brash, 1999). 12R-LOX and University Clinic, Heidelberg, Germany and 5Institute of Pathology, eLOX-3 differ from all other LOXs by unusual structural and Technische Universita¨tMu¨nchen, Munich, Germany enzymatic features. 12R-LOX represents the only mammalian 6These authors contributed equally to this work. LOX that forms products with R-chirality, and eLOX-3 Correspondence: Peter Krieg, Genome Modifications and Carcinogenesis, functions as a hydroperoxide by exhibiting only German Cancer Research Center, Im Neuenheimer Feld 280,D-69120 latent activity (Kinzig et al., 1999; Krieg et al., Heidelberg, Germany. E-mail: [email protected] 1999; Yu et al., 2003; Zheng and Brash, 2010). Both Abbreviations: ARCI, autosomal recessive congenital ichthyosis; CE, cornified envelope; CLE, corneocyte lipid envelope; eLOX-3, epidermal LOX-3; EOS, act in sequence to convert fatty acid substrates via R- esterified omega-hydroxyacyl-sphingosine; HETE, hydroxyeicosatetraenoic hydroperoxides to specific epoxyalcohol derivatives. With acid; HPETE, hydroperoxyeicosatetraenoic acid; HPLC, high-performance as substrate, the 12R-LOX/eLOX-3 pathway liquid chromatography; LOX, lipoxygenase; MS/MS, tandem mass results in the production of a specific R-epoxyalcohol of the spectrometry; OS, omega-hydroxyacyl-sphingosine; TEWL, transepidermal water loss; TLC, thin-layer chromatography hepoxilin family, members of which are known to function as Received 28 April 2012; revised 6 June 2012; accepted 24 June 2012; signaling molecules (Brash et al., 2007). Recent studies published online 26 July 2012 suggested a structural role of the 12R-LOX/eLOX-3 pathway

172 Journal of Investigative Dermatology (2013), Volume 133 & 2013 The Society for Investigative Dermatology P Krieg et al. Aloxe3 Knockout Mice

ac 100 +/+ E16.5 +/– 95 –/–

90 Body weight (%) Body weight 85 0 60 120 180 240 E17.5 Time (min)

b 8.0 ****

) **** –2 6.0 m –1 4.0 E18.5

2.0 TEWL (g h 0.0 +/+ +/– –/– +/+ +/– –/–

Figure 1. Loss of epidermal barrier function in epidermal lipoxygenase-3 (eLOX-3)-deficient mice. (a) Dehydration over time. Data are presented as percentage of initial body weight of wild-type ( þ / þ ), heterozygous ( þ /), and homozygous (/) mutant mice (n ¼ 5 for each genotype). Only homozygous mutants showed a weight loss of B3% per hour. (b) Transepidermal water loss (TEWL) measured on the lateral surface of newborn wild-type ( þ / þ , n ¼ 9), heterozygous ( þ /, n ¼ 18), and homozygous (/, n ¼ 16) mutant mice. (c) Barrier-dependent dye exclusion assay on wild-type and eLOX-3 mutant mice at the ages indicated. In control embryos, a decrease of skin permeability was observed from E16.5 to E18.5 as the barrier is acquired in the dorsal to ventral pattern, whereas eLOX-3-knockout embryos remained completely stained. ****Pp0.0001. via processing of o-hydroxy acylceramides as a prerequisite RESULTS for the formation of the corneocyte lipid envelope (CLE) Postnatal death and loss of skin barrier function in mice (Zheng et al., 2011). lacking eLOX-3 Mutations in ALOX12B or ALOXE3 have been detected in For targeting the Aloxe3 , we chose a conditional gene- B15% of ARCI patients and represent the second most targeting approach utilizing the Cre-loxP and flp-FRT common cause of ARCI (Eckl et al., 2009; Fischer, 2009). systems. A targeting construct was prepared and eLOX-3- Studies of our group and others demonstrated that such null mice were generated as described in Supplementary mutations may completely abolish the catalytic activity of the Figure S1a online and under Supplementary Materials and respective LOX enzymes (Eckl et al., 2005; Yu et al., 2005). Methods online. The genotypes of the eLOX-3 null mice were In previous studies, we and others showed that inactiva- determined by PCR and Southern blot analysis (Supplemen- tion of 12R-LOX in mice leads to neonatal death due to a severely impaired permeability barrier function (Epp et al., tary Figure S1a–c online). RT-PCR, western blot analysis, and 2007; Moran et al., 2007). Our studies showed that the immunofluorescence revealed no expression of eLOX-3 wild- defective barrier function of neonatal 12R-LOX-knockout type mRNA and the absence of eLOX-3 protein in eLOX-3/ skin is associated with ultrastructural anomalies in the upper mice (Supplementary Figure S1d and e online, Figure 3). granular layer, impaired processing of profilaggrin to filaggrin On the basis of proposed close functional relationship of monomers, and disordered composition of -bound both enzymes in barrier formation, we expected the eLOX-3- ceramide species. The mature 12R-LOX-deficient mouse deficient mice to show a similar lethal phenotype as observed skin, as analyzed in skin grafts, developed an ichthyosiform earlier in12R-LOX-knockout mice (Epp et al., 2007). Whereas phenotype that closely resembled the phenotype observed in heterozygous animals (eLOX-3 þ /) were phenotypically ichthyosis patients (de Juanes et al., 2009). indistinguishable from wild-type mice and reproduced In this study, we generated and analyzed Aloxe3-null mice normally, eLOX-3/ mice were born at the expected that display a lethal phenotype associated with severe skin Mendelian ratio, but died within 5–12hours after birth owing barrier dysfunction, which is similar to that observed in 12R- to defective skin barrier function. Weighing revealed that LOX-knockout mice. The pivotal role of the 12R-LOX/eLOX- eLOX-3/ mice lost around 2.5% of body weight per hour, 3 pathway for the establishment of the epidermal barrier whereas their heterozygous and wild-type littermates function can be traced back to the oxygenation of linoleate- maintained weight (Figure 1a). Dehydration through the containing ceramides, which constitutes an important step in skin was assessed by measuring transepidermal water loss the construction of CLE and barrier formation. Moreover, the (TEWL). Homozygous mutant mice showed approximately 4- data indicate an additional role of eLOX-3 in epidermal fold increased TEWL compared with control littermates signaling. (Figure 1b).

www.jidonline.org 173 P Krieg et al. Aloxe3 Knockout Mice

We assessed outside-in barrier acquisition during ab embryonic development by performing a whole-mount skin toluidine blue penetration assay (Hardman et al., 1998). In wild-type animals, the staining reflects a dorsal to ventral pattern of decreasing skin permeability with proceeding barrier development between embryonic day 16.5 and 17.5, coming to a halt on embryonic day 18.5 when the skin has become completely impermeable. Skin of homozygous µ µ mutant mice, in contrast, remained permeable as indicated 100 m 100 m by intense staining (Figure 1c). We also investigated the barrier function of tight junctions by injecting biotin cd subcutaneously into newborn mice and measuring its diffusion into the epidermis. Prevention of diffusion was observed in the upper granular cell layer of the skin of both homozygous mutant mice and wild-type mice, indicating that eLOX-3 deficiency did not affect the barrier function of tight junctions (not shown).

Structural abnormalities of the skin of eLOX-3-deficient mice 2,500 nm 2,500 nm The skin of eLOX-3/ newborns closely resembled the phenotype observed earlier in 12R-LOX-knockout mice. The ef stratum corneum appeared to be more tightly packed than in control skin (Figure 2a and b), whereas no significant differences were observed in the basal, spinous, or granular layers. At the ultrastructural level, electron microscopy revealed hyperkeratosis of the skin of eLOX-3-deficient mice as indicated by the presence of very compact horny layers (Figure 2d). Furthermore, we noticed an increased occur- rence of transition cells underlying the stratum corneum and occasionally irregular tubulus-like structures in the 2,500 nm 250 nm horny layers, indicating defects or delay of the desquama- tion process in the skin of eLOX-3-knockout animals Figure 2. Structural skin abnormalities of epidermal lipoxygenase-3 (eLOX-3)-deficient mice. Representative skin sections from (a) newborn (Figure 2e and f). The characteristic vacuole-like structures / observed earlier in the granular layers of neonatal 12R- wild-type and (b) eLOX-3 littermates stained with hematoxylin–eosin. The staining shows a more tightly packed stratum cormeum but otherwise LOX-deficient skin, however, could not be detected in the normally stratified organization of in knockout mouse skin. / skin of eLOX-3 mice. Electron microscopy of skin from (c) control and (d–f) eLOX-3/ neonates. (d) Stratum corneum of mutant skin. Note the markedly piled-up cornified Epidermal differentiation cells. (e) Accumulation of transition cells (asterisk) that underlay the stratum To unveil defects in epidermal differentiation, we studied corneum. (f) Irregular tubular-like structures in the horny lamellae (arrow) the distribution as well as the expression levels of may indicate defects or delay of the desquamation process. Bars ¼ 100 mm for a and b; 2,500 nm for c–e; and 250 nm for f. differentiation markers. Immunofluorescence staining was performed on the back skin of newborn mice using the following markers: keratin 14, keratin 10, keratin 6, filaggrin, involucrin, loricrin, and repetin. No obvious alterations Altered lipid composition of the expression pattern could be detected for keratins As lipid analysis of the skin from 12R-LOX-deficient mice had 10 and 14, involucrin, or repetin. Keratin 6 staining, revealed a significant decrease of protein-bound o-hydro- however, which was found to be restricted to hair follicles xyceramides, whereas other lipid components had remained in wild-type mice, was strongly increased in basal keratino- unchanged, we assumed that loss of these covalently bound cytes of eLOX-3-deficient animals (Figure 3). The near- ceramide species, which are the major constituents of the normal expression of the early and late differentiation CLE and thus essential for permeability barrier function, markers was confirmed by western blot analysis, which would be mainly responsible for epidermal barrier failure. also indicated that processing of profilaggrin to filaggrin Bringing forward a new model of the assumed role of the monomers was not disturbed in eLOX-3-knockout skin—in 12R-LOX/eLOX-3 pathway in skin barrier function, Brash and contrast to neonatal 12R-LOX-deficient skin that completely colleagues recently suggested that oxygenation of the lacks mature filaggrin monomers (not shown). In summary, linoleate-containing esterified omega-hydroxyacyl-sphingo- except for keratin 6 none of the early, intermediate, and late sine (EOS) ceramides by 12R-LOX and eLOX-3 is required to differentiation markers analyzed were altered in eLOX-3- facilitate hydrolysis and subsequent covalent linkage of the deficient mice. o-hydroxyceramides to proteins of the cornified envelope

174 Journal of Investigative Dermatology (2013), Volume 133 P Krieg et al. Aloxe3 Knockout Mice

s.c.

s.g. s.s. s.b. eLOX-3 eLOX-3 +/+ –/–

s.c. s.g. s.s. 12R-LOX s.b. 12R-LOX +/+ –/–

K6 K6 Flg Flg +/+ –/– +/+ –/–

K14 K14 IvI IvI +/+ –/– +/+ –/–

K10 K10 Rptn Rptn +/+ –/– +/+ –/–

Figure 3. Detection of epidermal lipoxygenase-3 (eLOX-3), 12R-LOX, and epidermal differentiation markers in the skin from control and eLOX-3-deficient mice. Skin sections from control and eLOX-3/ mice were immunostained for the indicated proteins. Nuclei are counterstained with Hoechst 33258. Specific staining (arrows) for eLOX-3 is observed in keratinocytes of the granular layers of wild-type mice but absent in mutants, whereas a similar staining pattern for 12R-LOX is unaffected. Note that except for keratin 6 (K6) where staining is seen in basal layers of mutant skin but restricted to hair follicles in wild-type mice, keratin 14 and 10 (K14, K10), filaggrin (Flg), involucrin (Ivl), and repetin (Rptn) staining is found at the typical localizations and at similar intensities in control and mutant skin. s.b., stratum basale; s.c., stratum corneum; s.g., stratum granulosum; s.s., stratum spinosum. Bars ¼ 20 mm.

(CE) (Zheng et al., 2011). To test this hypothesis in our levels of heterozygous animals did not differ significantly eLOX3-knockout model, we analyzed the ceramide compo- from those of wild-type mice. The observed difference sition of the epidermis of control and mutant mice by thin- between eLOX-3/ and 12R-LOX/ mice became more layer chromatography (TLC) and high-performance liquid evident when looking at different ceramide species that vary chromatography-tandem mass spectrometry (HPLC-MS/MS). in the length of their long-chain fatty acid (Figure 4d). In TLC analysis revealed no differences in the levels of free eLOX-3/ mice, levels of ceramides with fatty acids lipids (Figure 4a), whereas the subfractions of ester-bound containing 30 or 32 carbon atoms were not significantly ceramides differed significantly between control and knock- lower than in heterozygous or control animals, whereas 12R- out mice: they were strongly reduced in eLOX-3-deficient LOX-deficient mice showed a marked reduction of all OS epidermis but completely lost in 12R-LOX-knockout mice species investigated. (Figure 4b). We next evaluated the total amount of o- To clarify the impact of eLOX-3 on eicosanoid signaling, hydroxyacyl-sphingosine ceramides Cer[OS (18)] by HPLC- relevant free lipids of the epidermis of newborn mutant mice MS/MS (Figure 4c). Both in eLOX-3/ and 12R-LOX/ mice were further investigated by HPLC-MS/MS (Figure 4e and f). omega-hydroxyacyl-sphingosine (OS) formation was clearly Knockout of eLOX-3 had no detectable influence on the lower than in the wild type. However, 12R-LOX-deficient concentrations of the –derived LOX products, 9- animals showed a more distinct reduction. Cer[OS (18)] hydroxyoctadecadienoic acid and 13-hydroxyoctadecadienoic

www.jidonline.org 175 P Krieg et al. Aloxe3 Knockout Mice

a 12R-LOX eLOX-3 b

–/– +/+ –/– +/+

Fatty acids 12R-LOX eLOX-3

–/– +/– +/+ –/– +/– +/+ Chol Chol NS (24) OS OS AS (24) Protein-bound Free cd+/+ –/– eLOX-3 –/– 12R-LOX 1.5 1.0

1.0 ** ** *** ** 0.5 *** **

Cer [OS (18)] 0.5

Fold change ± SEM Fold *** Fold change ± SEM Fold *** *** *** *** *** *** *** 0.0 n=5 n=8 n=4 n=6 n=10 n=5 0.0 +/+ +/– –/– – – +/+ +/– –/– 30:0 32:0 32:1 34:0 34:1 36:0 36:1 eLOX-3 12R-LOX Cer [OS (18)]

ef eLOX-3 +/+ (n = 7) 60,000 +/– (n = 10) 2.5 50,000 12S–HETE –/– (n = 6) 12R–HETE 40,000 2.0 30,000 20,000

Intensity (c.p.s.) 10,000 Calibrator 1.5 0

60,000 1.0 50,000 Fold change ± SEM Fold 40,000 30,000 12S–HETE 0.5 20,000

Intensity (c.p.s.) 10,000 Epidermis 0.0 0 010203040 Time (min) 9-HODE15-HETE12-HETE 13-HODE Trioxilin A3Trioxilin B3 Hepoxilin B3

Figure 4. Lipid analysis of the epidermis of epidermal lipoxygenase-3/ (eLOX-3/) and 12R-LOX/ mice. (a, b) Thin-layer chromatography (TLC) analysis of free and ester-linked lipids from epidermis samples of newborn eLOX-3- or 12R-LOX-deficient and control mice. The ceramide nomenclatures used in this figure are as suggested previously (Motta et al., 1993). The positions of comigrating standards are indicated. (c, d) Protein-bound ceramides were investigated by high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS). Results were normalized to epidermal tissue weight and displayedin relation to the wild-type mice. (c) Peak areas of ceramide species were combined and normalized to epidermal weight. (d) Amounts of individual OS species that differ in the length of the fatty acid attached. (e, f) Levels of free LOX products in the epidermis of eLOX-3/ mice. (e) Relative amounts of different linoleic acid– and arachidonic acid–derived LOX products were determined by HPLC-MS/MS. The results were normalized to epidermal tissue weight. (f) Epidermis samples were further analyzed by chiral HPLC, which indicated that 12-hydroxyeicosatetraenoic acid (12-HETE) was exclusively in S-configuration. AS, ceramide AS; Chol, cholesterol; c.p.s, counts per second; HODE, hydroxyoctadecadienoic acid; NS, ceramide NS; OS, ceramide OS. **Pp0.01; ***Pp0.001.

acid, and the arachidonic acid–derived LOX product, 15- 2001), we speculated that 12-HETE detected in mouse skin hydroxyeicosatetraenoic acid (15-HETE), whereas higher is not a 12R-LOX product. Indeed, chiral analysis revealed amounts of 12-HETE were detected in eLOX-3/ epidermis this to be exclusively 12(S)-HETE, indicating that loss of than in the samples from heterozygous or wild-type mice eLOX-3 activity results in an accumulation of this eicosanoid (Figure 4e). As mouse 12R-LOX does not metabolize free derived from the 12(S)-LOX pathway (Figure 4f). The 12-LOX arachidonic acid, but only esterified substrates (Siebert et al., pathway is bifurcated at the level of its primary product

176 Journal of Investigative Dermatology (2013), Volume 133 P Krieg et al. Aloxe3 Knockout Mice

Bioactive lipids Structural lipids (signaling) (formation of CLE) ω Very long-chain fatty acid, -hydroxy (C28–36) H OH N H2C COOH O OH O Linoleic acid Sphingosine O Arachidonic acid EOS (ceramide 1)

12S-LOX 12R-LOX

H OH N COOH H2C O O OH OOH COOH O HOO 12S-HPETE 9R-HPODE-EOS OH eLOX-3 12S-HETE HOOC HOO H OH 9R-HPODE COOH N COOH HO H2C O O OH O HO O O O OH

Hepoxilin A3 Hepoxilin B3 9R(10R), 13R-EpHOE-EOS hydrolase Hydrolase

HOOC O OH COOH HO COOH 9R(10R), 13R-EpHOE HO OH HO HO HO NH H OH O 2 N C H2C O Trioxilin A3 Trioxilin B3 OH OH Gln ω-OH-OS CE proteins TGase 1

H OH N H2C O OH O O C Covalently bound OS Glu

Figure 5. Dual signaling and structural functions of epidermal lipoxygenase-3 (eLOX-3) in the mouse epidermis. Left panel: involvement of eLOX-3 in eicosanoid signaling. 12S-hydroperoxyeicosatetraenoic acid (12S-HPETE) generated via the 12S-LOX pathway is converted by eLOX-3 to hepoxilin A3 and hepoxilin B3, which are further metabolized to the corresponding trioxilins. Right panel: involvement of eLOX-3 in ceramide processing. The linoleate moiety of esterified omega-hydroxyacyl-sphingosine (EOS) ceramide is oxidized by 12R-LOX to the 9R-hydroperoxide derivative that is subsequently converted by eLOX- 3 to a 9R,10R epoxy-13R-hydroxy-epoxyalcohol derivative. LOX-mediated oxygenation of EOS is required to facilitate hydrolysis and subsequent covalent linkage of the o-hydroxyceramides to proteins of the cornified envelope (CE). The covalent coupling of OS ceramides as major constituents of the corneocyte lipid envelope (CLE) is an important step in establishing the water barrier. 9R(10)R, 13R-EpHOE, 9R(10R)-epoxy-13R-hydroxy-octadecaenoic acid; 12-HETE, 12-hydroxyeicosatetraenoic acid; HODE, hydroxyoctadecadienoic acid.

12-hydroperoxyeicosatetraenoic acid (12-HPETE), which can earlier for 12R-LOX-knockout mice (Epp et al., 2007), either be reduced to 12-HETE or converted to . indicated by weight loss (2.5% per hour vs 10%), increase Metabolites of this pathway such as hepoxilin B3,trioxilin in TEWL (4-fold vs 8-fold), and a maximum life span of about A3, and trioxilin B3 were found to be strongly reduced in 12 vs 3 hours. eLOX-3/ mice (Figure 6a). In 12R-LOX-deficient epidermis, The typical histological phenotype of skin from ichthyosis in contrast, no alterations in the levels of hepoxilin metabolites patients is characterized by variable epidermal hyperplasia were detected (data not shown). This indicates an important and marked hyperkeratosis, and is thought to result role of eLOX-3 in the respective hepoxilin signaling cascade from a physical compensation for the defective cutaneous and reveals an additional function of eLOX-3 in mammalian permeability barrier required for survival in a terrestrial skin as a hepoxilin synthase (Figure 5). environment. The macroscopic appearance of both 12R- LOX/ and eLOX-3/ mouse skin resembles that of mutant DISCUSSION human skin albeit at different degrees. Indeed, previous Loss-of-function mutations in the LOX genes ALOX12B and skin grafting studies confirmed that matured 12R-LOX- ALOXE3 are among the most frequent causes of ARCI. deficient skin developed an ichthyosiform phenotype similar Knockout of these LOX genes in the mouse models described to that seen in severe ichthyosis (de Juanes et al., 2009). here and in previous reports confirm a close functional Although our electron microscopy studies revealed less relationship of the corresponding LOX enzymes, 12R-LOX pronounced structural abnormalities in the skin of eLOX-3- and eLOX-3, and their essential role in skin barrier formation. deficient mice than in 12R-LOX/ skin, they clearly In both models, the mice die soon after birth owing to severe indicated the onset of compensatory hyperkeratosis and TEWL. Our data show that eLOX-3-deficient mice exhibit defects or delay of the desquamation process. Similarly, similar but somewhat less severe barrier defects than reported ectopic interfollicular expression of keratin 6 in neonatal

www.jidonline.org 177 P Krieg et al. Aloxe3 Knockout Mice

eLOX-3/ skin as shown by immunofluorescence may be unreported. Anton and Vila (2000) previously reported indicative of hyperproliferation. elevated levels of endogenous hepoxilins and trioxilins in This study provides insights into the mechanisms by which human psoriatic scales and the biosynthesis of hepoxilin-type loss of eLOX-3 function might affect epidermal barrier and products and their triol derivatives in isolated human skin homeostasis. Unlike all other members of the LOX epidermal fragments (Anton et al., 1998). family, eLOX-3 displays only latent dioxygenase activity but Although the exact steric identification of the hepoxilin exhibits a dominant hydroperoxide isomerase activity that metabolites in mouse skin remains to be established, the converts fatty acid hydroperoxides, the usual LOX products— hepoxilin B product had the same gas chromatography–mass preferentially products of 12R-LOX—to epoxyalcohols and spectrometry characteristics as the synthetic standard of . Although initially a function of the 12R-LOX/eLOX-3 10R-hydroxy-11S,12S-hepoxilin B3, which is the product sequence in eicosanoid signaling was hypothesized, of eLOX-3-catalyzed transformation of 12S-HPETE (Yu a recent study linked the 12R-LOX/eLOX-3 pathway to skin et al., 2003). Furthermore, reduction of hepoxilin metabo- ceramide processing and consequently to the formation of the lite levels in eLOX-3-deficient epidermis was accompanied CLE (Zheng et al., 2011). The study showed that linoleate- by increased levels of 12-HETE, which was shown by containing EOS ceramide is a natural substrate for the chiral analysis to be exclusively in the S-configuration. 12R-LOX/eLOX-3 pathway being oxygenated by consecutive Given the aforementioned inactivity of mouse 12R-LOX actions of 12R-LOX and eLOX-3 to a 9R-hydroperoxide- toward free arachidonic acid, we conclude that the and a corresponding 9R,10R-epoxy-13R-hydroxy-derivative. hepoxilin metabolites are derived from 12S-HPETE formed The model suggests that the LOX-catalyzed oxygenation by platelet-type 12S-LOX and/or epidermal 12S-LOX, both facilitates hydrolysis of the oxidized linoleate moiety result- of which are expressed in the mouse epidermis (Heidt ing in a free o-hydroxyl on the ceramides for coupling to the et al., 2000). cross-linked proteins of the CE. Cross-linked OS ceramides Hepoxilins and their triol derivatives have been shown to and o-hydroxy very long-chain fatty acids are the major havearoleinavarietyofphysiological processes, including constituents of the CLE that surrounds the corneocyte, and the inflammation and neurotransmission. Most functions appear to covalent coupling is an important step in sealing the water be mediated by influencing intracellular calcium concentrations barrier. (Pace-Asciak, 2009). In the skin, various potential mechanisms The results of our study are in line with the model of Zheng of action have been discussed for these metabolites including et al. and provide further evidence for a consecutive effects on a putative membrane receptor, ichthyin (NIPAL4), oxygenation of the linoleate esterified in EOS by 12R-LOX implicated through genetic studies in ichthyosis patients or and eLOX-3. Lipid analysis showed a significant decrease of through activation of PPAR receptors (Lefevre et al., 2004; Yu covalently bound OS ceramide species in eLOX-3-deficient et al., 2007). Further research is required to determine the exact skin, with a reduction of the total amount of bound ceramides mode of action and the physiological role of hepoxilin-type to half of the wild-type level, whereas in 12R-LOX-deficient metabolites in the epidermis. mice a reduction to less than 10% of the wild-type level was Taken together, our data indicate signaling and structural observed. The less pronounced loss of ester-bound ceramides functions of eLOX-3 in the epidermis, as summarized in in eLOX-3/ mice may be explained by a 12R-LOX- Figure 5. mediated partial oxidation of the linoleate moiety to a 9R- In addition to its function in the skin, eLOX-3 has also hydroperoxide derivative, which may permit a reduced been proposed to have a role in adipose tissue (Fu¨rstenberger hydrolysis of the linoleate and subsequent covalent linkage et al., 2007). This view is supported by our previous studies of resultant o-hydroxy ceramides, whereas interruption of the demonstrating that eLOX-3 is critically involved in adipocyte first oxidation step by 12R-LOX deficiency completely differentiation in vitro. The data suggest that eLOX-3, in prevents hydrolysis of the EOS ceramides. The loss of bound synergistic action with xanthine , produces ceramides is accompanied by barrier defects of increasing hepoxilins that function as endogenous PPARg activators severity in the different mouse models, indicating that loss of capable of promoting early steps in the conversion of ceramide binding may be the main cause of epidermal barrier preadipocytes into adipocytes (Hallenborg et al., 2010). In failure in the mutant mice. It remains to be elucidated this study, histological examination of homozygous and whether the free oxygenated cleavage products may have heterozygous eLOX-3-knockout mice showed normal organ signaling functions in barrier formation or other aspects of development compared with control mice (not shown). skin physiology beyond those hypothesized recently (Brash Furthermore, based on light microscopic inspection, no et al., 2007). alterations in brown and white adipose tissues could be Analysis of the free oxygenated fatty acid metabolites detected in neonatal mice, suggesting that eLOX-3 deficiency furthermore revealed strongly reduced levels of different had no effect on fetal adipose tissue development. A possible hepoxilin metabolites (the trihydroxy hydrolysis product of role of eLOX-3 in adipose tissue in vivo at later develop- the unstable hepoxilin A3 as well as hepoxilin B3 and its mental stages and in metabolism under distinct physiolo- corresponding triols) in eLOX-3-deficient skin, indicating an gical conditions, however, remains to be elucidated. The additional function of eLOX-3 as a hepoxilin synthase in availability of the floxed eLOX-3 mice generated in this mammalian skin. We report on the presence of hepoxilin and study will help establish conditional knockout models with trioxilin in mouse skin, which to our knowledge is previously inducible tissue-specific eLOX-3 inactivation for further

178 Journal of Investigative Dermatology (2013), Volume 133 P Krieg et al. Aloxe3 Knockout Mice

studies of the physiological role of eLOX-3 in the skin and Histological and immunofluorescence analysis adipose tissue. For light microscopic observation, samples were fixed with 4% In conclusion, this study provides evidence for a structural formalin in PBS for 24 hours, dehydrated in 70% ethanol, and then role of the 12R-LOX/eLOX-3 pathway in skin barrier embedded in paraffin. Five-micrometer sections were mounted on formation where LOX-catalyzed oxygenation of the linoleate slides, dewaxed, rehydrated, and stained with hematoxylin–eosin. moiety of ceramides constitutes an indispensable step for the Sections were embedded in mounting medium (DakoCytomation, covalent linkage of hydroxyl ceramides to proteins of the CE Hamburg, Germany) and examined with an Axioskop 2 with a Zeiss and subsequently for the formation of the CLE. Furthermore, 25 /0.8 Plan-Neofluar objective connected to an Axiocam camera we unveiled an additional function of eLOX-3 linked to its using the Axiovision software (release 3.2; Zeiss, Go¨ttingen, hepoxilin synthase activity and to the 12S-LOX pathway. The Germany). For immunofluorescence microscopy, 3-mm cryosections impact of this hepoxilin pathway on skin physiology remains were fixed in acetone for 10 minutes at 20 1C, permeabilized to be elucidated. with 0.5% Triton X-100 in PBS, blocked with 1% BSA in PBS for 20 minutes, and incubated with the primary antibody for 1 hour. MATERIALS AND METHODS After washing three times (10 minutes each) with PBS, samples were Generation and maintenance of eLOX-3-deficient mice incubated with a fluorescent dye–labeled secondary antibody and The eLOX-3-deficient mouse line was generated by targeted gene Hoechst 33258 for 30 minutes, and washed as described above. disruptionas described in Supplementary Materials and Methods Images were acquired using a LSM 700 confocal microscope with an online. The animal experiments were conducted in accordance with EC Plan-Neofluar 40 /1.30 Oil DIC objective lens (Zeiss), the the guidelines of the Arbeitsgemeinschaft der Tierschutzbeauftragten software ZEN 2009 (v5.5.0.375; scan parameters 1,024 1,024 in Baden-Wu¨rttemberg (Officials for Animal Welfare) and were Pixel, speed 2, line average 2), and Adobe Photoshop CS3. approved by the Regierungspra¨sidium Karlsruhe, Germany (Az. G-46/06). Functional analyses of the epidermal barrier To determine the rate of fluid loss, newborn animals were separated Antibodies from their mothers and put on a heating pad set at 37 1C. Body weight Rabbit polyclonal antipeptide antibodies detecting 12R-LOX and was monitored every 30 minutes until homozygous mutant mice eLOX-3, and mouse anti-12R-LOX mAb, have been described earlier died. The rate of TEWL from the skin of newborn mice was deter- (Epp et al., 2007). Rat anti-eLOX-3 mAbs were raised using mined with a Tewameter (Courage and Khazaka, Cologne, Germany). His-tagged eLOX-3 protein as immunogen. Other primary antibodies used were goat anti-GAPDH (Santa Cruz Biotechnology, Heidel- Toluidine blue staining of mouse embryos berg, Germany), rabbit anti-keratin 14, rabbit anti-keratin 10, rabbit The developmental stage of mouse embryos was determined based anti-keratin 6, rabbit anti-filaggrin, rabbit anti-involucrin (all Covance, on the assumption that fertilization occurred in the middle of the Princeton, NJ), and rabbit anti-repetin (gift from D Hohl). Secondary dark cycle the day before plugs were identified. The embryos were antibodies included Alexa Fluor 488 goat anti-mouse IgG (Invitrogen, subjected to methanol dehydration and subsequent rehydration as San Diego, CA), Alexa Fluor 488 goat anti-rat IgG (Invitrogen), CY3 described previously (Hardman et al., 1998), washed with PBS for goat anti-rabbit IgG (Jackson ImmunoResearch, Suffolk, UK), and 1 minute, and stained in 0.1% toluidine blue O/PBS for 30 minutes. horseradish peroxidase–conjugated anti-rabbit IgG (BD Biosciences, After decoloration in PBS for 15 minutes, the embryos were Palo Alto, CA). photographed.

Epidermal protein extraction and western blot analysis Electron microscopy Trunk epidermis of newborn mice was separated mechanically from All specimens were fixed in a solution of 3% glutaraldehyde in 0.1 M the dermis following submersion in hot phosphate-buffered saline cacodylate buffer, pH 7.4, for at least 2 hours at room temperature, (PBS) (56 1C) for 30 seconds and snap-frozen in liquid nitrogen. cut into pieces of B1mm3, washed in buffer, postfixed for 1 hour at Epidermal tissue was disintegrated for 30 seconds in a ball mill 4 1C in 1% osmium tetroxide, rinsed with water, dehydrated through (Micro-Dismembrator S, Sartorius, Go¨ttingen, Germany) at liquid graded ethanol solutions, transferred into propylene oxide, and nitrogen temperature and then homogenized in lysis buffer contain- embedded in epoxy resin (glycidether 100). Ultrathin sections ing 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 2 mM EDTA, were treated with uranyl acetate and lead citrate, and then examined 10% glycerol, 150 mM NaCl, 50 mM Tris (pH 7.4), and protein with an electron microscope (Philips EM 400, Eindhoven, 1 inhibitors (1 mM phenylmethylsulfonylfluorid, aprotinin 1 mgml The Netherlands). and leupeptin 1 mgml1). Western blot analysis was performed as previously described (Epp et al., 2007). Lipid analysis For the extraction of epidermal lipids and , epidermis Pathohistological evaluation samples were dissected and snap-frozen in liquid nitrogen. Neonatal mice were killed by decapitation, fixed in 4% para- Sphingolipids and eicosanoids were extracted according to Zheng formaldehyde, and embedded in paraffin blocks for histological et al. (2011). High-performance thin-layer chromatography separa- evaluation. One-micrometer sections were cut and stained with tion of lipids, HPLC-MS/MS, and LC-MS/MS analysis were performed hematoxylin–eosin. Images of the sections were taken at 20 as described earlier (Deems et al., 2007; Gelhaus et al., 2011; magnification with an automated slide scan microscope (dotSlide Jennemann et al., 2012) and in Supplementary Materials and System, Olympus, Hamburg, Germany). Methods online.

www.jidonline.org 179 P Krieg et al. Aloxe3 Knockout Mice

CONFLICT OF INTEREST proliferator-activated receptor gamma activity. MolCellBiol30: The authors state no conflict of interest. 4077–91 Hardman MJ, Sisi P, Banbury DN et al. (1998) Patterned acquisition of skin ACKNOWLEDGMENTS barrier function during development. Development 125:1541–52 We thank A Brash for generously providing ceramide standards and G Heidt M, Fu¨rstenberger G, Vogel S et al. (2000) Diversity of mouse Fu¨rstenberger for critical comments on the manuscript. Technical support by lipoxygenases: identification of a subfamily of epidermal isozymes the DKFZ Light Microscopy Facility is gratefully acknowledged. This work exhibiting a differentiation-dependent mRNA expression pattern. Lipids was supported by grants from the Deutsche Forschungsgemeinschaft (KR 905/ 35:701–7 6-2, KR 905/7-1, and SCHN 569/4-1). Jennemann R, Rabionet M, Gorgas K et al. (2012) Loss of ceramide synthase 3 causes lethal skin barrier disruption. Hum Mol Genet 21:586–608 SUPPLEMENTARY MATERIAL Kinzig A, Heidt M, Fu¨rstenberger G et al. (1999) cDNA cloning, genomic structure and chromosomal localization of a novel murine epidermis- Supplementary material is linked to the online version of the paper at http:// type lipoxygenase. Genomics 58:158–64 www.nature.com/jid Krieg P, Heidt M, Siebert M et al. (2002) Epidermis-type lipoxygenases. Adv Exp Med Biol 507:165–70 REFERENCES Krieg P, Siebert M, Kinzig A et al. (1999) Murine 12(R)-lipoxygenase: Anton R, Puig L, Esgleyes T et al. (1998) Occurrence of hepoxilins and functional expression, genomic structure and chromosomal localization. trioxilins in psoriatic lesions. J Invest Dermatol 110:303–10 FEBS Lett 446:142–8 Anton R, Vila L (2000) Stereoselective biosynthesis of hepoxilin B3 in human Lefevre C, Bouadjar B, Karaduman A et al. (2004) Mutations in ichthyin a new epidermis. J Invest Dermatol 114:554–9 gene on 5q33 in a new form of autosomal recessive Brash AR (1999) Lipoxygenases: occurrence, functions, catalysis, and congenital ichthyosis. Hum Mol Genet 13:2473–82 acquisition of substrate. J Biol Chem 274:23679–82 Moran JL, Qiu H, Turbe-Doan A et al. (2007) A mouse in the 12R- Brash AR, Yu Z, Boeglin WE et al. (2007) The hepoxilin connection in the lipoxygenase, Alox12b, disrupts formation of the epidermal permeability epidermis. FEBS J 274:3494–502 barrier. J Invest Dermatol 127:1893–7 Deems R, Buczynski MW, Bowers-Gentry R et al. (2007) Detection and Motta S, Monti M, Sesana S et al. (1993) Ceramide composition of the quantitation of eicosanoids via high performance liquid chromatography- psoriatic scale. Biochim Biophys Acta 1182:147–51 electrospray ionization-mass spectrometry. Methods Enzymol 432: Oji V, Traupe H (2006) Ichthyoses: differential diagnosis and molecular 59–82 genetics. Eur J Dermatol 16:349–59 de Juanes S, Epp N, Latzko S et al. (2009) Development of an ichthyosiform Pace-Asciak CR (2009) The hepoxilins and some analogues: a review of their phenotype in Alox12b-deficient mouse skin transplants. J Invest biology. Br J Pharmacol 158:972–81 Dermatol 129:1429–36 Siebert M, Krieg P, Lehmann WD et al. (2001) Enzymic characterization Eckl KM, de Juanes S, Kurtenbach J et al. (2009) Molecular analysis of 250 of epidermis-derived 12-lipoxygenase isoenzymes. Biochem J 355: patients with autosomal recessive congenital ichthyosis: evidence for 97–104 mutation hotspots in ALOXE3 and allelic heterogeneity in ALOX12B. J Invest Dermatol 129:1421–8 Yu Z, Schneider C, Boeglin WE et al. (2005) Mutations associated with a congenital form of ichthyosis (NCIE) inactivate the epidermal lipox- Eckl KM, Krieg P, Kuster W et al. (2005) Mutation spectrum and functional ygenases 12R-LOX and eLOX3. Biochim Biophys Acta 1686:3:238–47 analysis of epidermis-type lipoxygenases in patients with autosomal recessive congenital ichthyosis. Hum Mutat 26:351–61 Yu Z, Schneider C, Boeglin WE et al. (2007) Epidermal lipoxygenase products of the hepoxilin pathway selectively activate the nuclear receptor Epp N, Fu¨rstenberger G, Muller K et al. (2007) 12R-lipoxygenase deficiency PPARalpha. Lipids 42:491–7 disrupts epidermal barrier function. J Cell Biol 177:173–82 Yu Z, Schneider C, Boeglin WE et al. (2003) The lipoxygenase gene ALOXE3 Fischer J (2009) Autosomal recessive congenital ichthyosis. J Invest Dermatol implicated in skin differentiation encodes a hydroperoxide isomerase. 129:1319–21 Proc Natl Acad Sci USA 100:9162–7 Fu¨rstenberger G, Epp N, Eckl KM et al. (2007) Role of epidermis-type Zheng Y, Brash AR (2010) Dioxygenase activity of epidermal lipoxygenase-3 lipoxygenases for skin barrier function and adipocyte differentiation. unveiled: typical and atypical features of its catalytic activity with Prostaglandins Other Lipid Mediat 82:128–34 natural and synthetic polyunsaturated fatty acids. J Biol Chem 285: Gelhaus SL, Mesaros AC, Blair IA (2011) Cellular lipid extraction for targeted 39866–75 stable isotope dilution liquid chromatography-mass spectrometry Zheng Y, Yin H, Boeglin WE et al. (2011) Lipoxygenases mediate the effect of analysis. J Vis Exp 57, pii: 3399 in skin barrier formation: a proposed role in releasing Hallenborg P, Jorgensen C, Petersen RK et al. (2010) Epidermis-type omega-hydroxyceramide for construction of the corneocyte lipid lipoxygenase 3 regulates adipocyte differentiation and peroxisome envelope. J Biol Chem 286:24046–56

180 Journal of Investigative Dermatology (2013), Volume 133