The acyl-CoA binding protein is required for normal epidermal barrier function in mice.

The acyl-CoA binding protein (ACBP) is a 10 kDa intracellular protein expressed in all eukaryotic species. Mice with targeted disruption of Acbp (ACBP(-/-) mice) are viable and fertile but present a visible skin and fur phenotype characterized by greasy fur and development of alopecia and scaling with age. Morphology and development of skin and appendages are normal in ACBP(-/-) mice; however, the stratum corneum display altered biophysical properties with reduced proton activity and decreased water content. Mass spectrometry analyses of lipids from epidermis and stratum corneum of ACBP(+/+) and ACBP(-/-) mice showed very similar composition, except for a significant and specific decrease in the very long chain free fatty acids (VLC-FFA) in stratum corneum of ACBP(-/-) mice. This finding indicates that ACBP is critically involved in the processes that lead to production of stratum corneum VLC-FFAs via complex phospholipids in the lamellar bodies. Importantly, we show that ACBP(-/-) mice display a ∼50% increased transepidermal water loss compared with ACBP(+/+) mice. Furthermore, skin and fur sebum monoalkyl diacylglycerol (MADAG) levels are significantly increased, suggesting that ACBP limits MADAG synthesis in sebaceous glands. In summary, our study shows that ACBP is required for production of VLC-FFA for stratum corneum and for maintaining normal epidermal barrier function.

investigate the biochemical changes responsible for the macroscopic phenotype.
Here we report for the fi rst time that specifi c targeting of the ACBP gene in mice leads to a signifi cant decrease in the very long chain free fatty acids (VLC-FFA) content and changes in the biophysical properties of the stratum corneum. Importantly, this situation leads to a signifi cantly compromised epidermal barrier function. Our results indicate that ACBP is involved in processes that lead to production of VLC-FFAs via complex phospholipids in the lamellar body lipids.

Generation of ACBP-defi cient mice
Mice with targeted disruption of the ACBP gene were generated as described ( 24,29 ). Mice carrying the targeted ACBP gene were identifi ed using primer pairs i ) 5 ′ -AGG ATC TCC TGT CAT CTC ACC TTG CTC CTG and 5 ′ -AAG AAC TCG TCA AGA AGG CGA TAG AAG GCG ( ‫ف‬ 500 bp fragment within the neo R cassette) and ii ) 5 ′ -AGG ATC TCC TGT CAT CTC ACC TTG CTC CTG and 5 ′ -GTA TCT GCT CAT CTA TTC GGC TTG G ( ‫ف‬ 1,200 bp fragment spanning the 3 ′ region of the neo R cassette to the 5 ′ region of intron 2). The ACBP +/+ allele was identifi ed using primers 5 ′ -GGG TCC GGG AAG GGT TGG AGC and 5 ′ -GGC GCT TCA CCT CCT CAG CGG (spanning ‫ف‬ 1,140 bp within the region 5 ′ of the initiator codon to the 5 ′ region within exon 2). Mice used for experiments within this work were 3-to 4-month-old mice from backcross generations 12-14 on the C57BL/6JBomTac genetic background. Mice were housed at the Biomedical Laboratory, University of Southern Denmark, under standard laboratory conditions including 12 h light/ dark cycle, free access to feed (altromin 1324 for maintenance, altromin 1314 for breeding and lactation) and water in a room with ‫ف‬ 55% relative humidity at 22°C ± 3°C. Breeding of transgenic mice and animal experiments were approved by the Danish Animal Experiment Inspectorate, and all animal experiments were conducted in conformity with the PHS policy on humane care and use of laboratory animals ( 30 ).
Studies with targeted disruption of the yeast ACBP (Acb1p) gene have shown that ACBP is important for synthesis of very long chain fatty acids (VLCFA) and sphingolipids ( 17 ). Furthermore, Acb1p-depleted cells show accumulation of autophagocytotic vesicles and multilobed vesicles and display a strongly perturbed plasma membrane structure, indicating that lack of ACBP severely affects vesicular traffi cking in yeast (17)(18)(19).
In mammalian cells, knockdown of ACBP has been reported to lead to a diminished potential of preadipocytes to undergo differentiation ( 20 ) and of hepatoma cells to express key enzymes in cholesterol and FA catabolic metabolism ( 21 ). Furthermore, reports have indicated that ACBP affects the transcriptional activity of nuclear hormone receptors by both interacting directly with the receptors ( 22 ) and by binding ligands ( 23 ).
Recently, we reported specifi c targeting of the ACBP gene in C57BL/6JBomTac mice ( 24 ). ACBP Ϫ / Ϫ mice are viable and fertile and are born in a normal Mendelian ratio. However, the ACBP Ϫ / Ϫ mice suffer from a poor adaptation to weaning and go through a crisis with decreased growth rate around that time. We have shown that the mice suffer from a delayed adaptation of the liver to weaning due to a delayed induction of lipogenic pathways. In contrast to our data, Kier and colleagues reported that targeted disruption of ACBP leads to embryonic preimplantation lethality ( 25 ). The reason for this is unclear, but it may be caused by differences in targeting strategies. Kier and coworkers deleted a large part of the Acbp proximal promoter region, which might have interfered with the regulation of other genes in the region, whereas our targeting approach involved only deletion of parts of introns 1 and 2 and exon 2, as described in Ref. 24 . During our studies, we also observed a pronounced skin and fur phenotype that is macroscopically similar to the phenotype of mice carrying a ‫ف‬ 400 kb pleiotropic deletion on chromosome 1, a region containing six genes, including the ACBP gene ( 26,27 ). The deletion was reported to increase lethality on the B6 background but not on a mixed 129/B6 background ( 28 ). The mutant mice on a mixed 129/B6 genetic background show sebocyte hyperplasia and sparse, matted, reddish hair with a greasy appearance. Examination of the fur lipid content revealed a decreased content of triacylglycerols and a concomitant increase of an unidentifi ed lipid species. Furthermore, hyperplasia of sebocytes in nose skin and skin from lower thorax was reported. Interestingly, ectopic expression of ACBP could at least partially rescue the skin phenotype of these mice, indicating that it is the deletion of the ACBP locus that causes the macroscopic changes in the skin and fur phenotype. Our observations in ACBP Ϫ / Ϫ mice confi rm this notion. To better understand how ACBP depletion affects skin and fur properties, we set out to Although our measurements are referenced to solutions displaying different pHs, we decided to correlate the quantity pH in Equation 2 with the activity of protons. This assumption take into account that the concentration of water in skin stratum corneum is far from being in excess, a requirement to use the term pH. Therefore, a decrease in the proton activity indicates that the measured apparent pH from Equation 2 increases (and vice versa).

Isolation of tissues for lipid extraction
Naked (Veet ® treated) skin from the belly was used for lipid analyses and the FA elongation assay. Tissue was left 4-5 days after hair removal prior to isolation for lipid analyses. For isolation of epidermis, tissue was left fl oating on 2.5 mg/ml Dispase II (Roche) in Hanks BSS (w/o phenol red, Gibco 14025) at 4°C over night. Thereafter, epidermis was gently lifted away from dermis (under a micro-dissection microscope), washed once in distilled water, and freeze-dried to determine dry tissue weight. Stratum corneum was isolated after overnight fl oating on 1% trypsin solution (cell culture grade). Cells from the viable epidermis were removed from stratum corneum sheets by repeated rigorous shaking. Each sample preparation was checked under the microscope to ensure lack of epidermal cells. Fur for lipid analyses was obtained from the body of mice by shaving with an ethanol-cleaned electrical shaver.

Quantifi cation of stratum corneum ceramide content by TLC
Ceramides were extracted as previously described ( 39 ) with slight modifi cations. Stratum corneum was extracted three times with n-hexan/ethanol 95:5 (v/v) under ultrasonication for 20 min. Pooled extracts from the same animal were fi ltered using a 0.45 m PTFE syringe fi lter (Carl Roth GmbH and Co. KG, pore size) and evaporated under N 2 . The residue was resolved in 1 ml chloroform/methanol 2:1 (v/v) and stored at Ϫ 30°C until analysis. Samples were analyzed using TLC. The preparation of the plate and the application of samples were carried out as previously described ( 40 ). Samples were run in duplicate with a ceramide mixture consisting of ceramide AP (L-and D-conformation), CerNS, CerNP, and CerEOS, applied in different amounts (20-800 ng) for calibration.

Quantifi cation of cholesterol and total FFA content by TLC
Epidermal and stratum corneum lipid extracts were diluted according to dry tissue weight, and lipid extracts corresponding to 100 µg starting material from each mouse were pooled. Aliquots of the pooled extracts were subjected to TLC. Lipid extracts were run parallel to 5 µg Nu-Chek 18-4A and either 5 µg cholesterol standard or 5 µg heptadecanoic acid standard. Plates were developed twice in petroleumether/diethylether/acetic acid 70/30/1 (v/v/v). For visualization of cholesterol, plates were stained with 10% FeCl 3 in 5% CH 3 COOH and 5% H 2 SO 4 (pink/purple for cholesterol) and baked at 100°C for ‫ف‬ 5 min. For FA visualization, plates were stained with 10% CuSO 4 in 8% H 2 PO 3 and were baked at 180°C for ‫ف‬ 8 min.

Two-photon excitation microscopy of excised mouse ear skin -LAURDAN GP studies
Mice were euthanized, and 3 µl 0.1 mM LAURDAN in DMSO were placed on the surface of the inside of the excised ear. Tissue was left incubating for ‫ف‬ 30 min at room temperature in a slightly humidifi ed chamber. Cover slips were wetted with water vapor to enhance contact with the tissue. LAURDAN generalized polarization (GP) measurements ( 32 ) were used to evaluate the organization of the extracellular lipids within stratum corneum, as described by ( 33 ). LAURDAN is an amphiphilic probe that preferentially partitions into lipid membranes. The fl uorescence emission properties are sensitive to the water dipolar relaxation process that occurs in the probe's environment. Briefl y, the energy of the LAURDAN excited singlet state progressively decreases when the extent of dipolar relaxation process is augmented, red shifting the probe's emission spectrum, i.e., decreasing the GP function to lower values (see Equation 1 ). The extent of water dipolar relaxation is normally related to the lipid packing of the studied membrane being low when the membrane packing is tight (e.g., membranes in the gel phase) ( 34 ). LAURDAN GP measurements were performed using the microscopy instrumental setup described in Ref. 35 . Likewise, LAURDAN GPs were calculated using the formula : The correction factor G was calculated by acquiring GP images of a known LAURDAN reference solution in the microscope at the same instrumental conditions used in the tissue sections (for further detail, see Ref. 35 ). LAURDAN (2 µM) in DMSO was used, with the predetermined GP = 0.011. The reference's GP was measured as described Ref. 35 .

BCECF fl uorescence lifetime imaging microscopy on fresh mouse ear skin
Mouse ear was incubated with 3 µl 0.1 mM BCECF in ethanol for 1 h at room temperature in vivo. Labeling was repeated every 15 min during the 1 h incubation time. Mice were then euthanized, and the ear was immediately covered with a water vaporwetted cover slip to enhance contact with the tissue. Lifetime of the probe in the stratum corneum was determined to evaluate the local proton activity in the tissue, as described by Ref. 36 . Measurements were performed using the same multiphoton excitation microscope described above. Fluorescence lifetimes were measured in the frequency domain using a fi eld programmable gate array (FPGA) card for detection. The FPGA and the microscope are controlled by software developed by the Laboratory for Fluorescence Dynamics ( 37 ). As a lifetime reference, we used 5 mM fl uorescein in ammonium acetate solution pH 9.2 (lifetime 4.05 ns). Excitation wavelength was 820nm and emission light was collected through a 525 ± 25 nm fi lter. Solutions experiments were carried out to validate measurement and calculation methods (10 µM BCECF in 0.2M Na 2 HPO 4 /0.1M citrate with 3 mM KCl/140 mM NaCl). BCECF lifetime and species fractions were determined using the phasor plot in the SimFCS software ( 38 ). Fractional intensities were determined from solutions containing fully protonated or deprotonated BCECF (pH below 4 and above 8, respectively). An indication of the local proton activity in the skin was then estimated from the lifetime images as described by was isolated and lipids extracted as described above. Extracted lipids were resuspended in synthetic lower phase containing 50 mg/l butylated hydroxyl toluene, according to wet weight of epidermis. Lipid extracts were hydrolyzed and FAs were converted to FA methyl esters (FAME) by incubation with 1ml 2.5% H 2 SO 4 in water-free methanol (SUPELCO) for 5 h at 80°C. FAMEs were extracted by addition of hexane and water followed by centrifugation. The upper phase was isolated and dried under a steam of N 2 . FAMEs were resuspended in chloroform and separated by reverse-phase TLC (KC18 TLC PLATES (Whatman, Ref. 45 ) as previously described ( 46 ). The radiolabeled FAMEs were detected using a Molecular Dynamics Storage phosphor screen and Typhoon Trio Scanner (Amersham Biosciences) and quantifi ed by ImageQuant 5.0. The detected FAMEs were normalized to [1][2][3][4][5][6][7][8][9][10][11][12][13][14] C]oleic acid content.

Analysis of 1-O -alkyl-2,3-diacylglycerols and triacylglycerides by mass spectrometry
Fur lipid extracts were dried under a stream of nitrogen and dissolved in 7.5 mM ammonium acetate in chloroform/ methanol/2-propanol (1:2:4, v/v/v). Fur lipid mass spectrometric analysis was performed on a hybrid LTQ Orbitrap mass spectrometer (Thermo Fisher Scientifi c, Bremen, Germany) equipped with a robotic nanofl ow ion source TriVersa (Advion BioSciences Ltd, Ithaca NY) using chips with 4.1 mm nozzle diameter. The ion source was controlled by Chipsoft 6.4 software (Advion BioSciences) and operated at the ionization voltage of 0.96 kV and gas pressure 1.25 psi. Plates with lipid extracts were chilled to 4°C. MS survey scans were acquired in positive ion mode using the Orbitrap analyzer operated under the target mass resolution of 100,000 (full width at half maximum, FWHM), defi ned at m/z 400 under automatic gain control set to 1e5 as the target value. Epidermal and stratum corneum lipid extracts were dried under a stream of nitrogen, dissolved in chlorofom:methanol 1:2 with 5% isopropanol and 5 mM acetic acid and analyzed on a QSTAR Hybrid LC/MS/MS Quadrupole TOF mass spectrometer. Injection was through NanoES capillaries, Proxeon, Medium (cat. no. ES380), samples were ionized by 900V, and data were acquired during 2 min with scan range m/z 150-1,400. The monoalkyl-diacylglycerol and triacylglycerol content was quantifi ed against the triheptadecanoin internal standard.

Transepidermal water loss measurements
Hair was removed from the abdominal skin of the mice by electrical shaving followed by 150 s treatment with Veet ® hair removal cream. Throughout the experiment, standard wooden bedding was replaced with Kleenex ® tissues. Transepidermal water loss (TEWL) was determined using a DermaLab ® TEWL probe (Cortex Technology, Denmark) in accordance with the guidelines of the Standardization Group of the European Dermatitis Society ( 47 ). Skin conductance was measured using a Der-maLab ® MOIST probe. Three-to four-month-old male mice were anesthetized with IsoFlo ® Vet (2.75% in 350ml/O 2 /min, Orion Pharma Animal Health, Denmark), and measurements were conducted on naked skin on the abdomen of the mice. TEWL data were collected when mean TEWL and MOIST values were equal within groups (Student t -test) for two consecutive days after ( у 5 days) Veet ® treatment (data not shown). Furthermore, mice were only included in the experiment when no damage was visible on the skin (redness, edema, irritation, scaling). To avoid any effect of the anesthetics on the TEWL values, TEWL measurements were initiated у 6 min after induction of anesthesia. During measurements, mice were placed on a temperature-controlled electrical blanket to avoid hypothermia.

Lipid extraction of samples for mass spectrometry analyses
Following addition of internal standard (1 nmol/mg epidermis and stratum corneum dry weight N-18:0 Phytosphingosine or 2nmol/mg fur triheptadecanoin) lipids were extracted by a modifi ed Bligh and Dyer ( 41 ) method. Dry tissue was homogenized in a total volume of 1.9 ml chloroform:methanol:0.88% KCl aq 2:1:0.8 (v/v/v). Following homogenization, the ratio of solvents was changed to chloroform:methanol:0.88% KCl aq 1:1:0.9 (v/v/v), and samples were mixed rigorously. Following centrifugation, the lower phase was transferred to a new tube, and the upper phase reextracted twice with synthetic lower phase. The combined lower phases were washed once with synthetic upper phase and dried under a stream of nitrogen. Lipids were resuspended in synthetic lower phase and fi ltered through a solvent resistant 0.45 µm fi lter prior to analysis. Ions in the effl uent were ionized by electrospray ionization with an electrode potential of 3,500 V and the masses of negative ions were detected by a Bruker Esquire-LC quadrupole ion trap mass spectrometer. This method was used for quantifi cation against the internal ceramide standard of epidermal and stratum corneum ceramides as well as stratum corneum free FAs.

Quantifi cation of epidermal VLCFA content by mass spectrometry
For quantifi cation of epidermal free FAs, one fourth of the extracted epidermal lipids were derivatized to acyl-choline esters essentially as described in Ref. 42

FA elongation activity assay
Full thickness hairless skin explants from the abdominal region were isolated and incubated in 10 mM EDTA in Dulbecco's PBS without calcium and magnesium. Following 1 h preincubation, 7 Ci/ml [1][2][3][4][5][6][7][8][9][10][11][12][13][14] C]oleic acid (PerkinElmer) complexed to BSA (1:4, BSA:FA) was added to each explant and incubation was continued for an additional 4 h. Explants were then transferred to ice cold 10 mM EDTA in PBS to stop the reaction ( 44 ). Epidermis Histological examination of the skin from ACBP Ϫ / Ϫ mice and ACBP +/+ controls revealed that there are no morphological differences between genotypes in either the thick epidermis of the foot ( Fig. 2D, E ) or the skin from the back ( Fig. 2F, G ). We have examined the abundance and appearance of sebocytes and pilosebaceous units during skin and hair development and did not observe developmental defects or other differences in sebocytes or pilosebaceous units between adult ACBP +/+ and ACBP Ϫ / Ϫ mice (data not shown). We furthermore examined the expression of the classical differentiation markers keratin 5, keratin 10, and loricrin in thin (back skin) and thick (foot sole) epidermis and found no differential expression or morphological differences in the epidermis when comparing ACBP Ϫ / Ϫ mice and ACBP +/+ controls ( Fig. 3 ). Keratin 5 expression was confi ned to the basal undifferentiated keratinocytes ( Fig. 3A, B ), whereas keratin 10 expression was seen throughout the differentiated live cell layers of epidermis ( Fig. 3C, D ) as well as stratum corneum. Loricrin expression was, as expected, highest in the stratum granulosum ( Fig. 3E, F ). Finally, ultrastructural studies using transmission electron microscopy did not reveal obvious defects in the ACBP Ϫ / Ϫ mouse skin (data not shown).

Stratum corneum lipid membranes from ACBP
؊ / ؊ mice show altered physical properties compared with ACBP +/+ mice To investigate whether disruption of the ACBP gene affected the biophysical properties of the skin, we performed a characterization of the skin surface by two-photon excitation microscopy. It has previously been reported that the lateral organization of stratum corneum lipid membranes would have pronounced consequences for the barrier capacity of the skin ( 49 ). Thus, we fi rst evaluated the lateral organization of lipid membranes existing in the skin stratum corneum by labeling excised mouse ear tissue with LAUR-DAN and evaluating it under a two-photon excitation microscope. The LAURDAN GP function is sensitive to the extent of water dipolar relaxation processes (and water content) occurring at the membrane interface where the probe is located ( 50 ). The LAURDAN GP values measured in ACBP +/+ and ACBP Ϫ / Ϫ mice indicate that in both cases the contributing membranes within the stratum corneum

Statistical analysis
Data are presented as mean ± SEM unless otherwise indicated. Data were analyzed by Student's t -test or one way ANOVA unless otherwise stated. Software for statistical analyses was SAS ® Statistical Analysis Systems (SAS) Analyst application (SAS release version 9. 1,[2002][2003][2004] by SAS Institute Incorporated, Cary, NC, and the GraphPad Prism software. Level of signifi cance was preset to 0.05.

RESULTS
Histological examination of the skin phenotype of ACBP ؊ / ؊ mice We previously reported the generation of mice with targeted disruption of ACBP ( 24 ). These mice are born in a normal Mendelian ratio and show similar fertility and lifespan as ACBP +/+ mice; however, they display a poor adaptation to weaning and need special care during that time ( 24 ). Furthermore, the mice develop a clear macroscopic skin and fur phenotype from ‫ف‬ 16 days of age ( Fig. 1A ). The fur becomes greasy and matted, and at the time of weaning, the ACBP Ϫ / Ϫ mice are easily distinguished from ACBP +/+ and ACBP +/ Ϫ littermates. The different appearance of the fur is sustained throughout life; however, the fur becomes less greasy and develops in some areas a reddish brown color ( Fig. 1B ). Furthermore, most mice develop alopecia and scaling of the skin in the naked areas with age ( Fig. 1C ). This phenotype is macroscopically similar to that of mice with the nm1054 mutation that deletes ACBP and at least fi ve other genes (26)(27)(28).
Evaluation of the relative expression of ACBP in the different layers of the skin showed that ACBP expression is highest in the suprabasal layers, especially stratum spinosum ( Fig. 2A ); however, expression is observed in all live cells in the epidermis ( Fig. 2A ). In sebocytes, we observed a low-to-intermediate staining, whereas keratinocytes along the hair shaft display more prominent staining ( Fig. 2B ). Sebocytes lining the sebaceous glands are more stained compared with the differentiated cells in the center of the gland ( Fig. 2B ). This pattern of expression in mouse skin is similar to that reported for human skin ( 48 ). Sections from ACBP Ϫ / Ϫ mice ( Fig. 2C ) as well as negative (IgG) controls (data not shown) showed no staining.

Local proton activity in ACBP ؊ / ؊ mouse stratum corneum is decreased
To further investigate the biophysical properties of the skin tissue, we examined the local proton activity of the skin outermost layer; i.e., the stratum corneum. Mouse ear skin was labeled with the BCECF fl uorescent probe, the fl uorescence lifetime of which is sensitive to the protonation of the probe ( 52 ). The apparent stratum corneum pH is increased from ‫ف‬ 5.5 ± 0.1, as observed in tissue from ACBP +/+ mice, to ‫ف‬ 5.9 ± 0.1 in tissue from ACBP Ϫ / Ϫ mice ( Fig. 5A ), with a slightly lower pH locally in the intercellular space ( ‫ف‬ 5.3 ± 0.1 in ACBP +/+ and ‫ف‬ 5.8 ± 0.1 in ACBP Ϫ / Ϫ mice, Fig. 5B ). These values are in region display tight lateral packing, i.e., corresponding to gel-like membranes in model systems (GP above 0.5, see Ref. 32 and references therein). However, a signifi cant increase in LAURDAN GP values in the stratum corneum membranes of ACBP Ϫ / Ϫ mice compared with ACBP +/+ controls is observed ( Fig. 4A ). This indicates that the water content of the stratum corneum extracellular lipid matrix in the ACBP Ϫ / Ϫ mice is lower than in ACBP +/+ controls. Consistent with previous observations ( 50,51 ) showing that membranes with a low extent of water dipolar relaxation processes display more narrow GP histograms, the LAURDAN GP histogram across each image is narrower in ACBP Ϫ / Ϫ compared with ACBP +/+ controls ( Fig. 4B , further illustrated by Fig. 4C, D ). FA chain length of the stratum corneum is decreased in the ACBP ؊ / ؊ mice As described previously, lateral organization of membranes composed of stratum corneum lipids is highly dependent on both pH ( 54 ) and lipid composition ( 55 ). We therefore determined the overall lipid composition of the stratum corneum and epidermis. By means of TLC, we detected no signifi cant differences between ACBP +/+ and ACBP Ϫ / Ϫ mice in overall FFA or cholesterol content in isolated stratum corneum and epidermis ( Fig. 6A , B ). In addition, we quantifi ed the content of ceramides by MS as well as TLC without detecting signifi cant quantitative or qualitative differences in stratum corneum or epidermis ( Fig. 6C and

supplementary Figs. I, A-E, and II, A-F).
To further investigate the composition of lipid subspecies, we determined the content of FFAs. The major FFA species in isolated stratum corneum are longer than 20 carbons in length, primarily C24 and C26 FAs ( 56,57 ), agreement with previous observations in mouse skin ( 53 ). Representative images are presented in Fig. 5C-F and show that the local proton activity is decreased in the skin in general as well as in the intercellular space. The alterations    Stratum corneum and epidermis w/o hairs were isolated from 3-to 4-month-old mice. Lipids were extracted, and lipids corresponding to 50 µg tissue dry weight per mouse were pooled for ACBP +/+ as well as ACBP Ϫ / Ϫ mice [n = 5 (epidermis) and n = 6 (stratum corneum) for both genotypes]. (A) Samples were analyzed by TLC in parallel with lipid standards (S1 and S2). Results are representative of at least three independent experiments. (B) Cholesterol and FFA content was semiquantifi ed using the gel-analyzing tool in Image J, and gray area of sample/gray area reference is shown. (C) For ceramide analyses, individual stratum corneum samples from 3-to 4-month-old mice of each genotype were analyzed by TLC, and the content of ceramides were quantifi ed as described in Experimental Procedures. Species type was assigned based on comigration with ceramide standards. Identifi cation of one ceramide species (substance 1) could not be done using the present ceramide standards. Data are presented as mean ± SD. n = 4-9 (ACBP Ϫ / Ϫ ), n = 5-10 (ACBP +/+ ). Ceramides are abbreviated according to three different fatty acid components (N, nonhydroxy fatty acid; A, ␣ -hydroxy fatty acid; EO, linoleic acid esterifi edhydroxy fatty acid) linked via an amide bond to the long chain base (S, sphingosine; P, phyto-sphingosine). D and L refer to D-and L-confi guration, respectively. CE, cholesterol ester; Chol, cholesterol; Epid, epidermis; Me-FA, methylated fatty acid; nn, unknown species. and derive primarily from hydrolysis of complex phospholipids in the lamellar bodies (58)(59)(60). Using MS, we were able to quantify the C20-C28 FFA content ( Fig. 7 ); however, for FAs shorter than C20 we had too high background contamination from plastics and solvents to quantify correctly; and for FAs longer than C28, the signal was too low and variable for quantifi cation. Interestingly, despite the lack of differences in the overall FFA content of the stratum corneum ( Fig. 6A ), we found a remarkable >50% decrease in the content of VLC-FFAs in stratum corneum of ACBP Ϫ / Ϫ mice compared with ACBP +/+ mice ( Fig. 7A ). Notably, the VLC-FFA profi le as well as the total VLCFA content in whole thickness epidermis was similar between genotypes ( Fig. 7B, C ), indicating that lack of ACBP specifi cally affects the FFA composition in the stratum corneum where the VLCFAs are more abundant ( Fig. 7A ). The only signifi cant difference in epidermal FFA content between ACBP +/+ and ACBP Ϫ / Ϫ mice is the increase in C20:1 FA ( Fig. 7B ), which may derive from MADAGs produced by the sebaceous glands ( 61,62 ). Because the disruption of ACBP specifi cally affected the abundance of VLCFA in the stratum corneum, we speculated that the abundant presence of ACBP in stratum spinosum may be a requirement for elongation of FAs to be used for establishment of the extracellular lipid lamellae in stratum corneum. To investigate whether disruption of ACBP affects FA elongation per se, we performed an elongation activity assay in skin explants by incubating these with radiolabeled FAs. However, we did not detect signifi cant differences in the elongation activity between ACBP +/+ and ACBP Ϫ / Ϫ skin explants ( Fig. 7D ).These results indicate that lack of ACBP perturbs the formation VLC-FFA by mechanism other than FA chain elongation.

The skin and hair lipidomes of ACBP ؊ / ؊ mice are enriched in monoalkyl-diacylglycerol species
To evaluate whether the sebum lipid composition was changed, thereby being a possible source of the increased content of C20:1FA in epidermis, we quantifi ed the content of neutral lipids within isolated stratum corneum, whole thickness epidermis, and fur. It became evident from TLC analyses that a lipid specie less polar than TAGs is highly abundant in the lipid extracts from stratum corneum, epidermis, and fur (data not shown). On the basis of its molecular weight as determined with four digits of precision of the m/z , we identifi ed this lipid as monoalkyldiacylglycerol (MADAG) using lipid MS (data not shown). As this species is particularly enriched in sebum that is secreted along the hairs ( 61 ), we included fur lipid extracts in the MS analyses of the MADAG and TAG content in the skin. Interestingly, in the stratum corneum and epidermis of ACBP Ϫ / Ϫ mice, the MADAG content is signifi cantly increased, while the content of TAGs is unchanged ( Fig. 8A ,  B ). In fur lipid extracts from ACBP Ϫ / Ϫ mice, we found signifi cantly increased amounts of MADAG lipids compared with extracts from ACBP +/+ mice, and concomitant with this, we observed a decreased content of TAGs ( Fig. 8C ). These results indicate that MADAG synthesis is significantly increased in the sebocytes, leading to dramatically ( Fig. 9 ). Similar results, but with higher experimental variation, were obtained on shaved, non-Veet ® -treated skin (data not shown). A similar trend, although not as pronounced, was observed in 3-week-old mice, indicating that impairment of the epidermal barrier is initiated at early age but worsens with age (mean TEWL ± SEM 4.77 ± 0.11 [ACBP +/+ , n = 10) and 5.18 ± 0.12 (ACBP Ϫ / Ϫ , n = 11) at age 3 weeks]. Thus, disruption of the ACBP gene signifi cantly impairs the epidermal permeability barrier in both young and adult mice, but the epidermal dysfunction is exacerbated in the adult mice.

DISCUSSION
In this article, we show that targeted disruption of ACBP in mice leads to a clearly distinguishable skin and fur phenotype with greasy and matted fur when the mice are approximately 16 days old. At older ages, the skin becomes dry and scaly, and some of the fur is lost. Careful histological examinations of skin from 3-week-old and 3-month-old mice show no gross differences in skin morphology, structure of sebaceous glands, or number of hair follicles between ACBP Ϫ / Ϫ and ACBP +/+ mice. Importantly, our data show for the fi rst time that the stratum corneum of increased levels of MADAGs in the skin and on the hair of ACBP Ϫ / Ϫ mice.

ACBP ؊ / ؊ mice display impaired epidermal permeability barrier function
There is substantial evidence that VLCFAs are important for the epidermal permeability barrier. First, investigations in reconstituted skin lipid mixtures have demonstrated that the FA chain length distribution dramatically infl uences lipid membrane lateral organization ( 55 ). Second, knockout of FA elongases expressed in the epidermis has been shown to cause impairment of the epidermal barrier (63)(64)(65)(66). Thus, the signifi cant decrease in VLC-FFA in the stratum corneum combined with the changes in the biophysical properties of the tissue prompted us to determine the rate of water loss across the skin as a measure of the permeability barrier function. To allow measurements on naked skin, hair was removed from the abdomen of 3-to 4-month-old mice by shaving followed by Veet ® treatment, after which the skin was allowed to recover for > 5 days until mean TEWL and MOIST values were equal within groups for two consecutive days (data not shown). Interestingly, ACBP Ϫ / Ϫ mice had a ‫ف‬ 50% increased water permeability compared with ACBP +/+ controls decreased amount of TAG and an increased amount of a nearly comigrating lipid species that they were unable to identify. This lipid was most likely MADAG, which we identifi ed in this study. In addition, dependent on the anatomical site of the biopsy, the nm1054 mutant mice were reported to have an increased number of sebocytes associated with most pilosebaceous units ( 26 ). We have carefully investigated the development, abundance, and appearance of pilosebaceous units in the skin from the back, but we did not detect any differences between ACBP +/+ and ACBP Ϫ / Ϫ adult mice. Additional phenotypic characteristics, such as hydrocephaly and anemia, were observed in the nm1054 mutant mice, and on a C57BL/6J background, the nm1054 mutation led to signifi cant prenatal lethality as well as increased lethality of newborns ( 28 ). On the basis of our results from careful characterization of ACBP Ϫ / Ϫ mice during both early backcross generations and in our present congenic strain (C57BL/6JBomTac), we conclude that these phenotypic characteristics are most likely related to the deletion of genes other than the ACBP gene. Consistent with the increased transepidermal water loss, the LAURDAN GP data indicate that there is a signifi cant decrease in the water content of ACBP Ϫ / Ϫ mouse stratum corneum membranes, as inferred from the higher LAUR-DAN GP values. Subtle changes in the permeability properties of the membranes participating in the barrier can be caused by lipid compositional changes. For example, it has been reported that the characteristic FFA composition of the stratum corneum membranes is a very important factor in promoting hydrocarbon chain mixing, defi ning lipid mixing properties and membrane stability ( 55,67 ). Similarly, the increased apparent pH detected in the stratum corneum of ACBP Ϫ / Ϫ mice may also modulate the stability of the membranes by enhancing ionization of the FAs. This will increase repulsion between polar head groups at the membrane interface, causing local instabilities that may lead to increased permeability across the membranes. This is consistent with the increased TEWL observed in ACBP Ϫ / Ϫ mice.
In terms of lipid composition, we did not detect major differences between ACBP +/+ and ACBP Ϫ / Ϫ mice in the total content of the classical skin lipids. Thus, cholesterol, overall FFA, and ceramide contents are not signifi cantly different between genotypes in either the stratum corneum or the epidermis. Interestingly however, the stratum corneum of ACBP Ϫ / Ϫ mice contains signifi cantly lower levels of several VLC-FFA species compared with ACBP +/+ mice. Our analyses allowed only for quantifi cation of FAs up to 28 carbons in length; however, it is likely that the content of FFAs with longer chain length is decreased as well. This decrease in VLC-FFA may be responsible for the changes in the physical properties (apparent pH, water content) within the stratum corneum lipid membranes.
Despite the decrease in VLC-FFA in stratum corneum, we did not detect differences in the elongation activity of epidermal explants, indicating that lack of ACBP does not affect FA elongation per se . ACBP may instead be required for incorporation of VLCFAs into complex lipids, such as phospholipids, that are excreted in the lipid lamellar bodies ACBP Ϫ / Ϫ mice display altered biophysical properties and a signifi cant decrease in VLC-FFA concomitant with a compromised epidermal barrier function. The phenotype of the ACBP Ϫ / Ϫ mice show several similarities to that of the previously reported nm1054 deletion mutants ( 26 ). These mice lack a ‫ف‬ 400 kb fragment of chromosome 1, including the Acbp locus and at least fi ve other genes (26)(27)(28), and were reported to have sparse, reddish, matted hair with a greasy appearance. Similar to our data from ACBP Ϫ / Ϫ mice, the lipids associated with the fur of nm1054 mutant mice were also different from the fur lipids of wild-type mice. Fleming and colleagues ( 26 ) showed a  that the TAG synthesis is decreased because synthesis of MADAG is so dramatically increased in the absence of ACBP. In any case, the quantitative increase and compositional change in fur lipids are likely to be responsible for the greasy appearance of the fur. Furthermore, although sebum lipids are not generally considered key actors in the skin barrier ( 68,70 ), it is possible that these lipids contribute to the altered biophysical properties of ACBP Ϫ / Ϫ mice stratum corneum.
In conclusion, the results reported here show that ACBP is required for normal epidermal barrier function. We demonstrate that lack of ACBP leads to signifi cantly decreased levels of VLCFA in the stratum corneum, despite the presence of normal elongation activity in whole thickness epidermis, indicating that ACBP plays a role in FA traffi cking in specifi c cell types in mammals, in this case the viable epidermis. The decrease in stratum corneum free FA chain length is likely to be directly related to the increased permeability of the epidermal barrier and hence the decreased polarity as well as the decreased local proton activity of the stratum corneum. Finally, our data support a role of ACBP in promoting TAG and limiting MADAG synthesis in the sebaceous glands. Future studies should address the exact biochemical role of ACBP for complex lipid synthesis in keratinocytes. and serve as the main source of extracellular VLC-FFA in stratum corneum (58)(59)(60)(68)(69)(70).
Decreased incorporation of either VLCFAs or linoleic acid into lamellar body lipids has been shown to compromise the epidermal permeability barrier in a number of mouse models. Interestingly, only a subset of these models suffer from a lethal phenotype, and for at least three of these, i.e. the acyl-CoA:diacylglycerol acyltransferase (DGAT) 2 ( 71 ), stearoyl CoA desaturase (SCD) 2 ( 72 ), and elongase of VLCFAs ELOVL 4 ( 63-65 ) knockout mouse models, the level of omega-hydroxylated ceramides is profoundly decreased in the skin. These mice die due to excessive dehydration within a few hours after birth. Contrary to these two mouse models, the ACBP Ϫ / Ϫ mice show normal levels of omega-hydroxylated ceramides in stratum corneum, which is consistent with the relatively mild barrier defect seen in the ACBP Ϫ / Ϫ mice.
In contrast to the lethal phenotype of DGAT2 Ϫ / Ϫ , SCD2 Ϫ / Ϫ , and ELOVL4 Ϫ / Ϫ mice, other models such as ELOVL3 Ϫ / Ϫ ( 66 ), SCD1 Ϫ / Ϫ ( 73 ), or mice with essential FA defi ciency (EFAD) (74)(75)(76), display a nonlethal phenotype with increased transepidermal water loss. Characteristic for these as well as for the ACBP Ϫ / Ϫ mice is the development of tousled fur, reduced fur content, change (browning) in coat color, and irritated, eczematous skin in adult mice ( 66 ). Despite the striking similarities between the skin and fur phenotypes of ACBP Ϫ / Ϫ and ELOVL3 Ϫ / Ϫ mice, we did not detect changes in epidermal FA elongation in ACBP Ϫ / Ϫ mice, suggesting that ACBP affects different biochemical steps of the synthesis of VLC-FFA for the stratum corneum. Consistent with an important role of ACBP in mouse skin, we show that ACBP is highly expressed in the suprabasal layers of epidermis and in the rim of the sebaceous glands. The expression profi le of ACBP in mouse epidermis is in agreement with previous observations in human skin ( 48 ). In relation to the observed fur phenotype of the ACBP Ϫ / Ϫ mice, we observed a signifi cant increase in MADAG levels in both epidermis and stratum corneum of ACBP Ϫ / Ϫ compared with ACBP +/+ mice. MADAGs found in the skin are primarily derived from sebocytes ( 61 ), and previous compositional characterization of skin surface lipids in mice have shown that FA C20:1 is the major FA in MADAGs produced by the sebaceous glands ( 61,62 ). These lipids are only in rare diseased cases produced by keratinocytes ( 77,78 ). We fi nd increased MADAG levels in epidermis, stratum corneum, and fur of ACBP Ϫ / Ϫ mice. We propose that as MADAGs contain abundant amounts of FA C20:1, it is likely this FA is generated as a side effect to the increased MADAG synthesis. However, we cannot exclude other origins of this FA or the MADAG species in epidermis.
Consistent with the sebum being the main source of MADAG, examination of the fur lipids, which are derived almost exclusively from lipid biosynthesis in the sebaceous glands, showed a marked increase in MADAG content concomitant with a ‫ف‬ 60% decrease in TAG content in ACBP Ϫ / Ϫ compared with ACBP +/+ mice. Thus, absence of ACBP appears to shift the balance to favor MADAG synthesis over TAG synthesis in the sebocytes. The reason for this shift in lipid classes is unknown; however, it is possible