Male BALB/c mice (n=32; 8 weeks of age; weight, ~25 g) were obtained from Xi’an Jiaotong University Animal Center (Xi’an, China). The mice were fed standard chow and had access to water ad libitum, and were caged in a controlled environment (12 h light/dark cycle; temperature, 20-25°C; humidity, 45-55%). The mice were acclimatized for 3 days prior to any experiments. All experimental procedures were performed in accordance with the ‘Principles of Laboratory Animal Care’ (NIH) and with the approval of the laboratory animal care committee of Xi’an Jiaotong University (no. XJTULAC2017-733).

The skin on the backs of the mice was shaved using an electric shaver. The mice were divided into two group: i) Treated with topical 0.1% ATRA cream (Chongqing Winbond Pharmaceutical Co., Ltd., Chongqing, China) twice a day; and ii) treated with an oil/water cream (vehicle control) twice a day. The Psoriasis Area and Severity Index (PASI) scoring system was used to assess the severity of inflammation on the back skin of the mice (21). The PASI is widely used in clinical practice. Erythema, scales and thickening are scored as follows: 0, absent; 1, slight; 2, mild; 3, obvious; and 4, very obvious. The sum of the scores of these three factors is an indicator of the severity of inflammation (score of 0-12).

After 5 days of treatment, the mice were anesthetized using pentobarbital at 50 mg/kg. The skin from the backs of the mice, including the dermis and subcutaneous tissues, was removed. The skin was washed with pre-cooled PBS, excess fat was removed, and the specimens were placed in liquid nitrogen for RNA extraction. Subsequently, the mice were sacrificed by cervical dislocation.

ATRA-treated and untreated mouse dorsal skin specimens (1×1.5 cm) were fixed in 4% paraformaldehyde for 24 h at 4°C, dehydrated and embedded in paraffin. The samples were sectioned at 7 µm and stained with hematoxylin for 5 min and eosin for 2 min at room temperature. The epidermal thickness, parakeratosis, spongiosis and degree of inflammation were evaluated. Spongiosis (presence of widened intracellular spaces between epidermal keratinocytes) and inflammation were evaluated on a scale of 0-4. Epidermal thickness and spongy edema were measured using the built-in measurement tool in the software ndp.view2 (Hamamatsu Photonics K.K., Hamamatsu, Japan; https://www.hamamatsu.com/jp/en/product/type/U12388-01.html; version 2.7). The expanded capillaries were counted at high magnification (×200) in 10 randomly selected fields under an Olympus BX51 light microscope equipped with a DP70 digital camera (Olympus Corporation, Tokyo, Japan). Incomplete keratinization was assessed as present or absent (22).

The specimens were cut into 1-cm blocks and dipped in ice-cold sodium cacodylate buffer solution containing 2.5% glutaraldehyde at 4°C for 24 h. The blocks were washed three times and treated with 1% osmium tetroxide at 4°C for 1 h. The samples were dehydrated using an ethanol series and embedded in Epon 812. Ultra-thin sections were stained with lead citrate for 10 min and uranyl acetate for 30 min at room temperature. Ultrastructural changes were observed by transmission electron microscopy (TEM) using a H7650 microscope (Hitachi, Ltd., Tokyo, Japan).

HaCaT cells (immortalized keratinocytes) were obtained from the Fourth Military Medical University (Shan’xi, China) and grown in RPMI-1640 (GE Healthcare, Chicago, IL, USA) with 10% fetal bovine serum (Life Technologies; Thermo Fisher Scientific, Inc., Waltham, MA, USA) and 1% penicillin-streptomycin (GE Healthcare). Short tandem repeat profiling was performed by Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd. (Shanghai, China) to validate the cell line. The cells were subcultured following dissociation with 0.25% trypsin/0.05% EDTA (1:1) and passaged at a ratio of 1:4 every 3 days.

HaCaT cells (5,000 cells/well) were incubated in 96-well plates for 48 h and treated by different concentrations (0.1, 0.5, 1, 5 and 10 µM) of ATRA at 37°C for a further 36 h. A volume of 20 µl MTT solution (5 mg/ml) was added to each well and the culture continued for 4 h. Subsequently, the medium was removed, and 150ul dimethyl sulfoxide (DMSO) was added to each well. The light absorption value of each well was measured at optical density 490 nm (570 nm) with a micro-plate reader. Following determination of the half-maximal inhibitory concentration using different concentrations of ATRA (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany), HaCaT cells were incubated with or without 1 µM ATRA for 36 h at 37°C, which was chosen for use in the subsequent experiments. Stock solutions of ATRA (0.01 M) were prepared in DMSO, stored in the dark at −20°C, and further diluted with RPMI-1640. The concentration of DMSO was 1‰.

Microarray analysis was used primarily to identify candidates that were later confirmed via reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and/or immunohistochemistry. The gene expression profiles in the skin of mice and in HaCaT cells treated or untreated with ATRA were compared using NimbleGen Gene Expression analysis (Nimblegen Systems Inc., Madison, WI, USA). The ds-cDNA samples were washed and labeled according to the Nimblegen Gene Expression Analysis protocol (Nimblegen Systems Inc.). A NanoDrop ND-1000 was used to quantify the cDNA following purification. Cy3 labeling was conducted using the NimbleGen One-Color DNA labeling kit (NimbleGen Systems, Inc.), according to the manufacturer’s protocol. Subsequently, 100 U Klenow fragment (New England Biolabs, Inc., Ipswich, MA, USA) and 100 pmol deoxynucleoside triphosphates were added and incubated at 37°C for 2 h. One-tenth volume of 0.5 M EDTA was added to stop the reaction. The labeled ds-cDNA was purified using isopropanol/ethanol precipitation. The microarrays were immersed in the NimbleGen hybridization buffer/hybridization component A, which was supplemented with 4 µg ds-cDNA with Cy3 labeling. The reaction system was kept in a hybridization chamber (Nimblegen Systems, Inc.) at 42°C for 4 h. The NimbleGen Wash Buffer kit (Nimblegen Systems, Inc.) was used to wash the microarrays in an environment without ozone. The Axon GenePix 4000B microarray scanner was used to scan the microarrays.

The RNAeasy Midi kit (Qiagen AB, Sollentuna, Sweden) was used to isolate the total mRNA, according to the manufacturer’s protocol. RNA (1 µg) was added in a 20-µl reaction system (PrimeScript RT Reagent Kit; Takara, Otsu, Japan) at 37°C for 15 min and 85°C for 5 sec for cDNA synthesis. The SYBR Premix Ex Taq II amplification kit (Perfect Real Time; Takara Biotechnology Co., Ltd., Dalian, China) was used for qPCR on a Bio-Rad IQ5 System (Bio-Rad Laboratories, Inc., Hercules, CA, USA). The conditions were: i) 95°C denaturation for 10 min; and ii) 40 cycles at 95°C for 10 sec and 60°C for 30 sec. The primers are listed in Table I. β-actin was used for normalization. The results were analyzed using the IQ5 software (version 2.5). Gene expression levels were calculated using the 2^−ΔΔCq method (23).

ATRA-treated and untreated HaCaT cells were lysed in radioimmunoprecipitation assay buffer [150 mM NaCl, 50 mM Tris-HCl (pH 8.0), 0.5% deoxycholate, 0.1% SDS, 1% Nonidet P-40 and protease inhibitor cocktail; Sigma-Aldrich; Merck KGaA] at 4°C for 1 h. Protein concentration was measured using a bicinchoninic acid protein assay kit (Beyotime Institute of Biotechnology, Haimen, China). Equal amounts of proteins (30 µg) were separated on 12% SDS-PAGE gels and transferred to PVDF membranes (EMD Millipore, Billerica, MA, USA). TBS buffer containing Tween-20 (TBST) and 5% non-fat milk was used to block non-specific binding at room temperature for 2 h. Following blocking, primary antibodies were incubated overnight at 4°C: Rabbit anti-human Claudin-1 (cat. no. 13050-1-AP; ProteinTech Group, Inc., Chicago, IL, USA; 1:200), goat anti-human Claudin-4 (cat. no. 17664; Santa Cruz Biotechnology, Inc., Dallas, TX, USA; 1:200) and mouse anti-human β-actin (cat. no. 47778; Santa Cruz Biotechnology, Inc.; 1:1,000). The membranes were washed three times with TBST for 5 min. Horseradish peroxidase (HRP)-conjugated Affinipure Rabbit Anti-Goat IgG (H+L) (cat. no. SA00001-2; ProteinTech Group, Inc.; 1:3,000), HRP-conjugated Affinipure Goat Anti-Mouse IgG (H+L) (cat. no. SA00001-1; ProteinTech Group, Inc.; 1:3,000) and HRP-conjugated Affinipure Donkey Anti-Goat IgG (H+L) (cat. no. SA00001-3; ProteinTech Group, Inc.; 1:4,000) were incubated for 1 h at 37°C prior to visualization with enhanced chemiluminescence (Amersham; GE Healthcare). Densitometry was used to quantify the signal intensities using Quantity One 4.5 software (Bio-Rad Laboratories, Inc.). All measurements were performed in triplicates from three independent experiments.

HaCaT cells and cryostat sections (4 µm) of skin from ATRA-treated and untreated mice were fixed in ice-cold 4% paraformaldehyde in PBS for 30 min. Non-specific binding was blocked with 10% normal goat serum (OriGene Technologies, Inc., Beijing, China) for 20 min at 37°C. Goat anti-human Claudin-4 (1:50) and rabbit anti-human Claudin-1 antibodies (1:50) were incubated overnight at 4°C. PBS was used as a negative control. The sections were washed twice for 5 min in PBS. Alexa Fluor 594-labeled secondary antibodies (cat. nos. A32758 and A32740; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA; 1:200) were incubated for 1 h at room temperature and revealed using DAPI (1:5,000 in PBS) for 2 min at room temperature. A laser scanning confocal microscope (LSM510; Zeiss AG, Oberkochen, Germany) was used to capture images of the cells (×200 magnification).

All statistical analyses were conducted using SPSS 16.0 (SPSS, Inc., Chicago, IL, USA). Data are expressed as the mean ± standard deviation. All measurements were performed in triplicate from three independent experiments. The distribution was tested for normality using the Kolmogorov-Smirnov test. Statistical significance was evaluated by independent sample t-test for normally distributed data and using the Mann-Whitney U test for non-normally distributed data. P<0.05 was considered to indicate a statistically significant difference.

For the microarray data, the NimbleScan software (version 2.5; NimbleGen Systems, Inc.) was used to conduct grid alignment and analyze the expression profiles. The Robust Multichip Average (RMA) algorithm and quantile normalization in NimbleScan software were applied to the normalized expression data. The gene levels were inputted into the Agilent GeneSpring software (version 12.0; Agilent Technologies, Inc., Santa Clara, CA, USA). Fold-change filtering was applied for the identification of genes with differential expression levels. The threshold was set at-fold-change ≥2.0. The Agilent GeneSpring GX software (version 12.0; Agilent Technologies, Inc.) was used to conduct hierarchical clustering. The roles of the differentially expressed genes were determined using Gene Ontology (GO; www.geneontology.org) and pathway analyses (https://www.genome.jp/kegg) conducted on the basis of the standard enrichment computation method, and using Fisher’s exact test to determine whether the amount of overlaps between the GO annotation list and the list of differentially expressed genes was significant. P≤0.05 indicated that the GO term enrichment was statistically significant.

After 5 days of treatment, no marked alterations were detected in the control skin. Circumscribed erythema occurred after 3 days of ATRA application (n=6), with fine flat scales covering the surface of the erythema that peaked at 5 days (Fig. 1). Hematoxylin and eosin staining of ATRA-treated skin demonstrated crust, focal parakeratosis, hyperproliferation and intercellular edema of the stratum spinosum (Fig. 2A). In the ATRA-treated dermis, large amounts of inflammatory cell infiltration and capillary dilation were observed (Fig. 2A). Epidermal thickness was 2.5 times higher following ATRA intervention compared with that of the control group (P<0.0001), in addition to increases in parakeratosis (P<0.001) and spongiosis (P<0.002). In the dermis, perivascular lymphocytic infiltration (P<0.002) and telangiectasia (P<0.002) were observed in the dermal papilla and in the superficial layer (Fig. 2B).

Compared with control skin, keratohyalin granules were decreased in number in the SG (n=3; Fig. 3A). TEM also demonstrated that the keratinocyte cytoskeleton was damaged by ATRA and that the cytoskeletal network disappeared in the local upper stratum corneum (Fig. 3A). In the upper stratum corneum, multiple lipid droplets were observed in the cytoplasm of corneocytes in the ATRA-treated skin (Fig. 3B). In the epidermis of ATRA-treated skin, TEM revealed a disordered arrangement of stratum spinosum cells, in addition to the appearance of significantly larger nucleoli and wider intercellular spaces (Fig. 3B). Desmosomes were decreased and disordered in ATRA-treated skin (Fig. 3C). Quantitative analysis of keratohyalin granules, lipid droplets and desmosomes is presented in Fig. 3D.

With the aim of improving the understanding of the molecular mechanisms of ATRA-induced barrier dysfunction, gene expression array analysis (n=1) was used to identify candidate genes for barrier function in the mouse skin and HaCaT cells. There were 897 upregulated and 1,087 downregulated genes following treatment with ATRA in the mouse epidermis. Similarly, there were 1,220 upregulated and 905 downregulated genes following treatment with ATRA in HaCaT cells. The genes involved in epidermal barrier function were revealed by gene expression analyses and are presented in Tables II–V. The downregulated epidermal barrier-associated genes following treatment with ATRA in the mouse skin included genes involved in ‘cornified envelope components’, ‘intermediate filament’, ‘lipid metabolic process’, ‘gap junction’, ‘tight junction’ and ‘desmosome’ (Table II). The upregulated epidermal barrier-associated genes following treatment with ATRA in mouse skin included genes involved in ‘cornified envelope components’, ‘serine protease and protease inhibitors’, ‘lipid metabolic process’, ‘tight junction’ and ‘transcriptional factors’ (Table III). The downregulated epidermal barrier-associated genes following treatment with ATRA in HaCaT cells included genes involved in ‘intermediate filament’, ‘proteases and protease inhibitors’ and ‘tight junction’ (Table IV). The upregulated epidermal barrier-associated genes following treatment with ATRA in HaCaT cells included genes involved in ‘intermediate filament’, ‘proteases and protease inhibitors’ and ‘tight junction’ (Table V).

The pathway analyses demonstrated that differentially expressed genes were significantly associated with AJs, TJs and focal adhesion in ATRA-treated epidermal tissues. The top ten pathways of up- and downregulated differentially expressed genes are presented in Fig. 4A and B, respectively. The top ten three GO-fold enrichment of differentially expressed genes revealed significant downregulation of epidermal barrier function-associated processes, including ‘regulation of keratinocyte differentiation’, ‘keratin filament’ and ‘gap junction’ channel activity (Fig. 4C-H). In ATRA-treated epidermal tissues, a series of dysregulated genes associated with the epidermal differentiation complex (EDC) were observed, which included small proline-rich region proteins (SPRRs), filaggrin (FLG), loricrin (LOR) and the S100 gene family. These proteins are essential for the formation of the cell envelope during terminal differentiation. A number of these genes were members of the SPRR family: SPRR4 (−2.67-fold), SPRR2J (+2.38-fold) and SPRR2G (+3.57-fold). The remainder were part of the S100 gene family: S100A8 (+3.31-fold) S100A9 (+3.81-fold) and FLG (−2.13-fold) (Tables II and III). During keratinocyte differentiation, the late cornified envelope protein (LCE) was induced and had similar functions to the functions of the SPRR family (cross-linking proteins) in the CE. LCE1M (6.64-fold) was downregulated, while LCE3C (5.35-fold) and LCE3B (3.17-fold) were upregulated (Tables II and III). Nevertheless, no notable abnormalities were observed in the above genes in HaCaT cells. In ATRA-treated epidermal tissues and HaCaT cells, marked dysregulation of numerous keratins and keratin-associated proteins was observed, suggesting abnormal terminal differentiation of keratinocytes upon ATRA treatment. Serine proteases and serine protease inhibitors are involved in the differentiation of keratinocytes and desquamation (24). In ATRA-treated epidermal tissues and HaCaT cells, the differential expression of serine proteases and serine protease inhibitors was observed. Among them, it was observed that KLK6 and KLK10 were consistently expressed in the mouse chip and HaCaT cells, and all of them were upregulated.

Genes encoding enzymes associated with lipid metabolism were demonstrated to be induced by ATRA in the mouse epidermis. Compared with the cells, the mice exhibited significant downregulation of a subset of enzymes associated with ceramide, free fatty acids and phospholipid synthesis: i) Sphingomyelin phosphodiesterase 3 (SMPD3; -2.15-fold) and sphingomyelin phosphodiesterase 2 (SMPD2, -2.10-folds), which catalyze the conversion of sphingolipids to ceramides; ii) epidermal ceramide synthase 5 (LASS5/Cers5; -2.82-fold), which catalyzes the synthesis of ceramides; iii) lysophospholipase II (LYPLA; -2.00-fold), phospholipase A2 group IIE2 (PLA2G2E; -2.60-fold), and phospholipase C eta 2 (PLCH2; -2.13-fold), which catalyze free fatty acid formation from phospholipids; and iv) 1-acylglycerol-3-phosphate O-acyltransferase 4 (AGPAT4; -3.81-fold) and 1-acylglyc-erol-3-phosphate O-acyltransferase 5 (AGPAT5; -3.17-fold), which catalyze the acylation of lysophosphatidic acid into phosphatidic acid. The latter is the primary precursor of all types of glycerolipids. Furthermore, arachidonate lipoxygenase (ALOX12E; a member of the epidermis-type LOX family that catalyzes the conversion of polyunsaturated fatty acids into oxygenated products) was downregulated by ATRA. Conversely, the present results also demonstrated upregulation of different isoforms of the aforementioned enzymes (Table III).

The dysregulation of numerous cell-cell junction-associated genes was observed in ATRA-treated mouse skin. Among these genes were components of TJs (TJAP1, MPP7, MARVELD2 and CLDN2), gap junctions (GJA, GJA and GJB4) and desmosomes (DSG2). Compared with the mouse skin specimens, only the TJ protein molecules Claudin-4 and occludin were upregulated in the cells.

In ATRA-treated cells, CLDN1 was downregulated 0.66-fold, while CLDN4, TJP3 and GJB4 were upregulated 1.82-, 3.63- and 1.92-fold (Fig. 5A). In mice treated with ATRA, CLDN1 (0.8-fold) and FLG (0.61-fold) were downregulated, while CLDN4 (1.47-fold) and CLDN2 (2.49-fold) were upregulated (Fig. 5B). Western blot analysis of CLDN4 and CLDN1 revealed similar results to those of the RT-qPCR (Fig. 5C).

Immunofluorescence revealed that the intensity of Claudin-1 was not only reduced at the cell-cell contact areas, but also appeared to be discontinuous along the cell membranes of HaCaT cells treated with ATRA (Fig. 6). However, Claudin-4 staining was slightly increased and displayed a string-like localization pattern (Fig. 6).

A marked reduction in the staining intensity and localization of Claudin-1 in the upper epidermal layer of skin treated with ATRA was discovered compared with control skin (n=3; Fig. 7). By contrast, Claudin-4 staining was stronger in the ATRA-treated skin compared with control skin (Fig. 7).