Truth Seeker
New Member
ABSTRACT
We had previously shown that both locally produced endocannabinoids and exocannabinoids, via cannabinoid receptor-1 (CB1), are powerful inhibitors of human hair growth. To further investigate the role of the cannabinoid system in pilosebaceous unit biology, we have explored in the current study whether and how endocannabinoids have an impact on human sebaceous gland biology, using human SZ95 sebocytes as cell culture model. Here, we provide the first evidence that SZ95 sebocytes express CB2 but not CB1. Also, prototypic endocannabinoids (arachidonoyl ethanolamide/anandamide, 2-arachidonoyl glycerol) are present in SZ95 sebocytes and dose-dependently induce lipid production and (chiefly apoptosis-driven) cell death. Endocannabinoids also up-regulate the expression of key genes involved in lipid synthesis (e.g., PPAR transcription factors and some of their target genes). These actions are selectively mediated by CB2-coupled signaling involving the MAPK pathway, as revealed by specific agonists/antagonists and by RNA interference. Because cells with "silenced" CB2 exhibited significantly suppressed basal lipid production, our results collectively suggest that human sebocytes utilize a paracrine-autocrine, endogenously active, CB2-mediated endocannabinoid signaling system for positively regulating lipid production and cell death. CB2 antagonists or agonists therefore deserve to be explored in the management of skin disorders characterized by sebaceous gland dysfunctions (e.g., acne vulgaris, seborrhea, dry skin).–Dobrosi, N., Tóth, B. I., Nagy, G., Dózsa, A., Géczy, T., Nagy, L., Zouboulis, C. C., Paus, R., Kovács, L., BÃró, T. Endocannabinoids enhance lipid synthesis and apoptosis of human sebocytes via cannabinoid receptor-2-mediated signaling.
INTRODUCTION
THE ENDOCANNABINOID SYSTEM (ECS), that is, endocannabinoids (such as arachidonoyl ethanolamide, AEA, and 2-arachidonoyl glycerol, 2-AG), specific cannabinoid receptors (CB1 and CB2), and enzymes involved in the synthesis and degradation of endocannabinoids, has emerged as a versatile modulatory system, implicated in a plethora of physiological and pathophysiological regulatory mechanisms (12 3) . Classically, CB1-mediated effects were mostly described in the central nervous system and were shown to involve regulating e.g., synaptic functions, memory, and motor learning (3 4 5 6) . Peripherally, the ECS has become implicated in regulation of e.g., immune and cardiovascular processes, apparently chiefly via CB2-coupled signaling (7 8 9) .
Recent intriguing findings, however, have also identified the functional existence of various members of the ECS on numerous other, previously unappreciated cell and tissue types. Among these, we have been particularly interested in human and murine skin and related cutaneous phenomena. Elements of the ECS were extensively documented in epidermal keratinocytes (10 11 12) . Moreover, cannabinoids were shown to suppress in vitro proliferation (and differentiation) of cultured epidermal keratinocytes (11 , 13) , similar to the effects of selective CB2 agonists on human coronary artery smooth muscle cells (14) , as well as the in vivo growth of murine skin tumors (15) and human melanomas (16) . In addition, using double CB1/CB2 gene-deficient mice, Karsak et al. (17) have elegantly demonstrated that endocannabinoids attenuate allergic contact dermatitis.
We have recently identified a CB1-mediated mechanism for nonclassical, peripheral tissue activities of endocannabinoids and exocannabinoids in the human system (18) . Namely, using organ-cultured human scalp hair follicles, we have shown that locally produced endocannabinoids (via CB1 that is expressed mainly on outer root sheath keratinocytes of the follicle) inhibit human hair growth and induce premature apoptosis-driven involution of this complex miniorgan (catagen).
Hair follicles most commonly are arranged in pilosebaceous units, which display another characteristic adnexal structure of mammalian skin, i.e., the sebaceous gland (reviewed in refs. 19 , 20 ). It is extensively documented that epithelial cells of the sebaceous gland, i.e., sebocytes, play a central role in the regulation of cutaneous lipid homeostasis and that pathological malfunctions of these cells may result in such common cutaneous diseases as acne vulgaris (21 22 2324) .
Of importance, sebaceous gland cells reportedly also show CB receptor immunoreactivity in situ (12) . However, apart from anecdotes about marijuana users often developing acne, direct evidence for the presence of functional ECS in sebaceous glands and a description of the potential effects of endocannabinoids on various biological processes of human sebocytes are lacking.
In the current study, we have therefore analyzed the presence and function of the ECS and related signaling mechanisms in human sebaceous gland-derived cells, using the SZ95 sebocyte cell line, one of the best-established human sebocyte cell culture models (25 26 27) . Specifically, we intended to clarify which CB receptors are expressed by human sebocytes, and whether prototypic endocannabinoids (AEA, 2-AG) can be detected in SZ95 sebocytes. Furthermore, we have evaluated the effects of endocannabinoids on defined sebocyte functions (e.g., lipid synthesis, cell growth and death, gene expression), using an array of cellular and molecular assays, and have defined the involvement of various intracellular signaling pathways in mediating the effects of endocannabinoids (by employing specific agonists, antagonists, and RNA interference).
Collectively, these studies provide the first evidence that human sebocytes selectively express functional CB2, contain key endocannabinoids, and utilize this endogenous cannabinoid signaling system for the autocrine and paracrine control of human sebocyte lipid production and death.
MATERIALS AND METHODS
Materials
AEA, 2-AG, arachidonyl-2-chloroethylamide (ACEA), AM-251, iodo-resiniferatoxin (I-RTX), and GF10203X were purchased from Sigma-Aldrich (Taufkirchen, Germany). JWH-015 and GW9662 were obtained from Cayman Chemical Company (Ann Arbor, MI, USA). PD098059 and wortmannin were obtained from Calbiochem (Nottingham, UK), and AM630 was from Tocris Bioscience (Ellisville, MO, USA).
Cell culturing
Human immortalized SZ95 sebocytes (25 26 27) , were cultured in Sebomed® basal medium (Biochrom, Berlin, Germany) supplemented with 10% fetal bovine serum (Invitrogen, Paisley, UK), 1 mM CaCl2, 5 ng/ml human recombinant epidermal growth factor (Sigma-Aldrich), 50 U/ml penicillin, and 50 µg/ml streptomycin (both from Biogal, Debrecen, Hungary).
Determination of endocannabinoid levels
The levels of endocannabinoids (AEA, 2-AG) were determined by liquid chromatography/in line mass spectrometry as described in our earlier reports (18 , 28) .
Immunocytochemistry
SZ95 sebocytes were fixed in acetone, permeabilized by 0.1% Triton-X-100 (Sigma-Aldrich) in phosphate-buffered saline (PBS), and then incubated with rabbit anti-CB1 or anti-CB2 primary antibodies (Cayman) for 60 min (dilution 1:200). Slides were then incubated with a goat anti-rabbit fluorescein-isothiocyanate (FITC) -conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA) (dilution 1:200), and nuclei were visualized using DAPI (Vector). Cells were examined on a Nikon Eclipse E600 fluorescent microscope (Nikon, Tokyo, Japan) (29) .
Immunohistochemistry
The study was approved by the Institutional Research Ethics Committee and adhered to Declaration of Helsinki guidelines. Normal female trunk skin samples, obtained during plastic surgery, were fixed in 4% formalin and embedded in paraffin, and 4-µm-thick sections were obtained. After antigen retrieval and blocking of the endogenous peroxidase activity, tissue sections were incubated with the above CB1 and CB2 primary antibodies (Cayman) (dilution 1:150 for CB1, 1:200 for CB2). Sections were then incubated with the EnVision+ System Labeled polymer-HRP Anti-Rabbit (Dako, Glostrup, Denmark) with 3,3'-diaminobenzidine (DAB) visualization techniques. Tissue samples were finally counterstained with hematoxylin (Sigma-Aldrich) and mounted in aqueous mounting medium (Dako) (18 , 29 , 30) .
In the course of the immunohistochemistry, numerous control experiments were performed. As "staining-negative" controls, the appropriate primary antibodies were either omitted from the procedure or were preabsorbed with synthetic blocking peptides (Cayman) (see Fig. 1 ). As "tissue-negative" controls, CB-labeling was performed on tissues not expressing CB1 (human mast cell line HMC-1) (31) or CB2 (HMC-1, human hair follicle) (18 , 31) . In addition, the specificity of CB receptor staining was also measured 1) on tissues recognized to be CB1 (HaCaT epidermal keratinocytes, hair follicle) (10 , 13 , 18) or CB2 (HaCaT keratinocytes, human peripheral monocytes, spleen) positive (11 , 18 , 32) (data not shown); or 2) by employing another set of antibodies against CB1 and CB2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The application of these latter primary antibodies resulted in identical staining patterns (data not shown).
Western blot analysis
Western immunoblotting was performed as described in our earlier reports (29 , 30 , 33) . In brief, cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to BioBond nitrocellulose membranes (Whatman, Maidstone, England), and then probed with anti-CB1 or anti-CB2 receptor antibodies (1:200, Cayman); with a rabbit antibody against the mitogen-activated protein kinase (MAPK) Erk-1/2 (Santa Cruz); or a mouse antibody recognizing the phosphorylated form of Erk-1/2 (pErk-1/2, Santa Cruz) (1:1500 dilution in both cases). Horseradish peroxidase—polymer-conjugated, respective anti-rabbit or anti-mouse IgG antibodies (Envision labeling, DAKO) were used as secondary antibodies, and the immunoreactive bands were visualized by SuperSignal West Pico Chemiluminescent Substrate-enhanced chemiluminescence (Pierce, Rockford, IL, USA). Immunoblots were then subjected to densitometric analysis using an Intelligent Dark Box (Fuji, Tokyo, Japan) and the Image Pro Plus 4.5.0 software (Media Cybernetics, Silver Spring, MD, USA). To assess equal loading, membranes were stripped and then reprobed with a rabbit cytochrome-C (Cyt-C) antibody (Santa Cruz) followed by a similar visualization procedure as described above.
Reverse transcriptase-polymerase chain reaction (RT-PCR)
The expression of CB1 and CB2 receptor mRNA was determined by semiquantitative RT-PCR, as we have described before (18 , 29 , 30) . In brief, isolated total RNA was reverse-transcribed into cDNA and then amplified on a GeneAmp PCR System 2400 DNA Thermal Cycler (Applied Biosystems, Foster City, CA, USA) using optimized thermal protocols. Primers were synthesized by Invitrogen (CB1, forward: CAAGCCCGCATGGACATTAGGTTA, CB1, reverse: TCCGAGTCCCCCATGCTGTTATC; CB2, forward: TCCCACTGATCCCCAATGACTACC, CB2, reverse: AGGATCTCGGGGCTTCTTCTTTTG; glyceraldehyde 3-phosphate dehydrogenase (GAPDH), forward: ATGGTGAAGGTCGGTGTGAAC, GAPDH, reverse: GCTGACAATCTTGAGGGAGT). PCR products were visualized on 1.5% agarose gel with ethidium bromide (0.5 mg/ml, Sigma-Aldrich) under UV, and the photographed bands were quantified by Image Pro Plus 4.5.0 software.
Quantitative real-time PCR
Quantitative PCR (Q-PCR) was performed on an ABI Prism 7000 sequence detection system (Applied Biosystems) using the 5' nuclease assay, as detailed in our previous report (18 , 29) . To detect the expression of genes involved in the regulation of lipid synthesis (see also Table 1 ), the following TaqMan primers and probes were used: peroxisome proliferator-activated receptor (PPAR)α, forward: CATTACGGAGTCCACGCGT, PPARα, reverse: ACCAGCTTGAGTCGAATCGTT; PPARα, probe FAM-AGGCTGCAAGGGCTTCTTTCGGCG-TAMRA; PPARδ, forward: AGCATCCTCACCGGCAAAG, PPARδ, reverse: CCACAATGTCTCGATGTCGTG; PPARδ, probe FAM-CAGCCACACGGCGCCCTTTG-TAMRA; PPARγ, forward: GATGACAGCGACTTGGCAA, PPARγ, reverse: CTTCAATGGGCTTCACATTCA; PPARγ, probe FAM-CAAACCTGGGCGGTCTCCACTGAG-TAMRA; adipose differentiation-related protein (ADRP), forward: TGACTGGCAGTGTGGAGAAGA, ADRP, reverse: ATCATCCGACTCCCCAAGA; ADRP, probe FAM-TCTGTGGTCAGTGGCAGCATTAACACA-TAMRA; PPARγ-regulated angiopoietin-related protein (PGAR), forward: TCCGCAGGGACAAGAACTG, PGAR, reverse: CGGAAGTACTGGCCGTTGA; PGAR, probe FAM-TTGGAATGGCTGCAGGTGCCA-TAMRA; and a predesigned assay for cyclooxygenase-2 (COX-2) (Applied Biosystems, assay ID: Hs00153133_m1). As internal controls, transcripts of human cyclophilin-A (forward: ACGGCGAGCCCTTGG, reverse: TTTCTGCTGTCTTTGGGACCT; probe FAM-CGCGTCTCCTTTGAGCTGTTTGCA-TAMRA) were determined.
RNA interference (RNAi)
SZ95 sebocytes were seeded in 6-well culture plates in SeboMed medium lacking antibiotics. At 50—70% confluence, medium was replaced by serum-free OptiMEM (Invitrogen), and cells were transfected with various CB2-specific Stealth RNAi oligonucleotides (ID: HSS102085, HSS102086, HSS102087, Invitrogen) (40 nM) using Lipofectamine 2000 transfection reagent (Invitrogen). For controls, RNAi Negative Control Duplexes (Scrambled RNAi, Invitrogen) and CB1-specific Stealth RNAi oligonucleotides (ID: HSS102082) were employed. Three hours after transfection, medium was replaced by complete Sebomed® medium, and cells were allowed to recover for 24 h. The efficacy of siRNA-driven knockdown was daily evaluated by RT-PCR and Western blot analysis for 4 days (34) .
Determination of intracellular lipids
For semiquantitative detection of sebaceous lipids, SZ95 sebocytes were cultured on glass coverslips and treated with several compounds (AEA, 2-AG) for 24—48 h. Cells were fixed in 4% paraformaldehyde (Sigma-Aldrich), washed in 60% isopropanol, and stained in Oil Red O solution (0.3% in isopropanol) (Sigma-Aldrich). Cells were counterstained with hematoxylin (Sigma-Aldrich) and were mounted in aqueous mounting medium (Dako) (35 , 36) .
For quantitative measurement, SZ95 sebocytes (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One, Frickenhausen, Germany) in quadruplicates and were treated with compounds for 24—48 h. Supernatants were then discarded, and 100 µl of 1 µg/ml Nile red (Sigma-Aldrich) solution was added to each well. The emitted fluorescence was measured on a Molecular Devices FlexStation 384II Fluorescence Image Microplate Reader (FLIPR, Molecular Devices, San Francisco, CA, USA). Results are presented as percentages of the relative fluorescence units (RFU) in comparison with the controls, using 485-nm excitation and 565-nm emission wavelengths for neutral lipids, and 540-nm excitation and 620-nm emission wavelengths for polar lipids (35 3637) .
Determination of viable cell number
The number of viable cells was determined by measuring the conversion of the tetrazolium salt MTT (Sigma-Aldrich) to formazan by mitochondrial dehydrogenases. Cells were plated in 96-well plates (15,000 cells/well) in quadruplicates and were cultured for 24—48 h. Cells were then incubated with 0.5 mg/ml MTT, and the concentration of formazan crystals (as an indicator of the viable cell number) was determined colorimetrically (A550), according to our previous reports (29 , 33 , 34 , 38) .
Determination of apoptosis
Abolishment of mitochondrial membrane potential is one of the earliest markers of apoptosis (39) . To assess the process, mitochondrial membrane potential of SZ95 sebocytes was determined using a MitoProbe DilC1(5) Assay Kit (Invitrogen). Cells (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) in quadruplicate and were treated with various compounds for 24—48 h. After removal of supernatants, cells were incubated with DilC1(5) working solution (30 µl/well), and the fluorescence of DilC1(5) was measured at 630-nm excitation and 670-nm emission wavelengths using FLIPR (36) .
In addition, another hallmark of apoptosis (i.e., membrane perturbation) was also assessed by flow cytometry according to our previous reports (29 , 34 , 36 , 38) . In brief, following treatment with various agents, SZ95 sebocytes were harvested and stained with an Annexin-V-FITC/propidium iodide (PI) apoptosis kit (Sigma-Aldrich) following the manufacturer's protocol. Fluorescence intensity was measured by a Coulter Epics XL (Beckman Coulter, Fullerton, CA, USA) flow cytometer.
Determination of necrosis/cytotoxicity
Necrotic cell death was first determined by measuring the glucose-6-phosphate-dehydrogenase (G6PD) release (G6PD Release Assay Kit, Invitrogen). The enzyme activity was detected by a 2-step enzymatic process that leads to the reduction of resazurin into red-fluorescent resorufin. SZ95 sebocytes (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) in quadruplicate and treated with various compounds for 24—48 h. A 2x reaction medium was then prepared according to the manufacturer's protocol and was added to the wells in 1:1 dilution. The fluorescence emission of resorufin was monitored by FLIPR at 545-nm excitation and 590-nm emission wavelengths (36) .
The cytotoxic effects of cannabinoid treatment were also determined by Sytox Green staining (Invitrogen). The dye is able to penetrate (and then bind to the nucleic acids) only to necrotic cells with ruptured plasma membranes, whereas healthy cells with intact surface membranes show negligible Sytox Green staining. SZ95 sebocytes were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) and were treated with various compounds (AEA, 2-AG, JWH-015, ACEA) for 24—48 h. Supernatants were then discarded, and the cells were incubated with 1 µM Sytox Green solution. The fluorescence of Sytox Green was measured at 490-nm excitation and 520-nm emission wavelengths using FLIPR (36) .
Statistical analysis
When applicable, data were analyzed using a 2-tailed unpaired t test, and values of P < 0.05 were regarded as significant. In addition, statistical differences were further verified using 1-way ANOVA with Bonferroni and Dunnett post hoc probes, resulting in similar results (data not shown).
RESULTS
Human sebaceous gland epithelium and human SZ95 sebocytes express CB receptors
First, we intended to identify the existence of CB receptors in human sebaceous gland in situ and on human SZ95 sebocytes. Similar to a previous report (12) , in situ, both CB1- and CB2-like immunoreactivity was identified in the sebaceous gland epithelium of normal human scalp skin sections (Fig. 1A, B ). However, using mutually complementary and confirmatory immunocytochemistry and Western blot analysis on human SZ95 sebocytes, we were able to identify only the relatively high expression of CB2, whereas the appearance of CB1-like immunoreactivity, in all cases, was around the detection limit (Fig. 1C-E ). In addition, supporting the above findings, transcription of the CB2 (but not of CB1) gene in SZ95 sebocytes was demonstrated by RT-PCR (Fig. 1F ) and by real-time Q-PCR (not shown). For positive controls, human organ-cultured hair follicles (for CB1) (18) and human peripheral monocytes (for CB2) (32) (as well as other tissue and cell types listed in Materials and Methods, data not shown) were employed. Therefore, in striking contrast to human hair follicle epithelium in situ (18) , human sebocytes express primarily, if not exclusively, CB2, at least in vitro.
Human SZ95 sebocytes produce endocannabinoids
We also tested whether SZ95 sebocytes synthesize endocannabinoids. Using mass spectrometry, we were able to show that SZ95 sebocytes express both AEA (66.7±10 fmol/mg tissue) and 2-AG (6.2±1 pmol/mg tissue) (means±SE, n=4) at levels similar to those detected earlier in various skin samples; e.g., ∼50 fmol/mg tissue AEA in rat paw and mouse ear samples (17 , 40) or 20—30 pmol/mg tissue 2-AG in mouse ear skin (17) .
Endocannabinoids enhance lipid synthesis and induce apoptosis in SZ95 sebocytes
To further explore the functionality and the biological consequences of CB stimulation as well as possible auto- and paracrine-signaling events, we next investigated the effects of these endocannabinoids found in human SZ95 sebocytes on key functions of these cells. As revealed by Oil Red-O staining and quantitative Nile Red-based fluorescence assay, both AEA and 2-AG markedly (P<0.05) and dose dependently enhanced neutral lipid accumulation in SZ95 sebocytes (Fig. 2A, B ), reflecting stimulation of sebocyte differentiation (35) .
By MTT assay, we also showed that stimulation with these endocannabinoids decreased sebocyte viability (Fig. 2C ). To assess whether this effect was due to apoptosis and/or necrosis, first, flow cytometry analysis was performed. As seen in Fig. 2D , both AEA and 2-AG markedly (P<0.05) increased the number of Annexin-V-positive cells (reflecting phosphatidyl-serine translocation) (29 , 34 , 36 , 38 , 41) , whereas the number of double Annexin-V- and PI-positive SZ95 sebocytes was only slightly elevated. This suggests that exogenously applied endocannabinoids primarily induce sebocyte apoptosis. In further support of this concept, in a series of quantitative fluorimetric assays, AEA (as well as 2-AG, data not shown) significantly decreased mitochondrial membrane potential (another hallmark of apoptosis) (36 , 39) in a dose-dependent fashion (P<0.05), while only the highest concentration of AEA was able to moderately (yet significantly, P<0.05) increase Sytox Green accumulation and G6PD release, two complementary indicators of necrosis/cytotoxicity (Fig. 2C ) (36) .
The effects of AEA on human sebocytes are selectively mediated by CB2
We then investigated whether the cellular actions of AEA on SZ95 sebocytes were mediated by the CB receptors. As seen in Fig. 3A, B , the synthetic CB2-specific agonist JWH-015 (3 , 7 , 42) –but, notably, not the CB1-specific agonist ACEA (3 , 7 , 43) –mimicked the action of endocannabinoids to enhance lipid synthesis and to induce (chiefly apoptosis-driven) cell death. Moreover, the effects of AEA were prevented by the CB2-specific antagonist AM-630 (3 , 7 , 11 , 28) (Fig. 3C, D ) but not by the CB1-specific inhibitor AM-251 (3 , 7 , 18) (Fig. 3C ). These data suggested that the sebocyte-modulatory effects of endocannabinoids are mediated by CB2 but not by CB1.
However, AEA, in various cellular systems, may also act on another receptor, i.e., transient receptor potential vanilloid-1 (TRPV1, the capsaicin receptor) (3 , 44 , 45) . Furthermore, we have recently described the functional existence of TRPV1-mediated signaling both in human organ-cultured hair follicles (29) and in SZ95 sebocytes (36) . Therefore, to further dissect the exact cellular mechanisms of action of the endocannabinoid, we also measured the effect of the TRPV1 antagonist I-RTX (46) on the actions of AEA. As seen in Fig. 3C , I-RTX did not interfere with the ability of AEA to enhance lipid synthesis (and to induce cell death, data not shown).
Collectively, these functional data are in agreement with the apparent lack of CB1 expression in SZ95 sebocytes at either the protein or gene level (Fig. 1) and suggest that the cellular actions of endocannabinoids are mediated by CB2. To further assess the role of CB2, a series of RNAi experiments against the receptors was carried out (Fig. 4A, B ). Western blot and RT-PCR analysis revealed that the expression of CB2 was significantly knocked down by all 3 RNAi probes at day 2 after transfection and remained suppressed also on day 3. However, this phenomenon was reversible, because we observed a return of the immunosignals at day 4. Scrambled RNAi probes (Fig. 4A ) or RNAi oligonucleotides against CB1 (data not shown) had no effect on the expression of CB2, indicating the specificity of the CB2 knockdown.
We then investigated the effects of AEA-treatment (24 h) on the lipid synthesis of RNAi-transfected SZ95 sebocytes on day 2. Similar to the effects of CB inhibitors, RNAi knockdown of CB2 resulted in the loss of effect of AEA in enhancing lipid synthesis. In contrast, treatment with the CB1-targeted RNAi did not affect the cellular action of AEA (Fig. 4C ). Intriguingly, in SZ95 sebocytes with RNAi-mediated knockdown of CB2, we found markedly and significantly decreased basal lipid content as well (Fig. 4C ). This strongly suggests that CB2-mediated signaling indeed plays an important endogenous role in the constitutive regulation of sebocyte lipid synthesis.
The CB2-mediated cellular signaling involves the MAPK pathway
On various cell types, CB receptor-mediated signaling, initiated by either endocannabinoids or exocannabinoids, recruits multiple intracellular pathway systems, such as protein kinase C (PKC), MAPK, or phosphatidyl-inositol-3-kinase (PI3K) (47 48 49) . Therefore, we have also undertaken attempts to elucidate selected components of CB2-mediated intracellular signaling in human sebocytes.
As seen in Fig. 5A , the enhancement of SZ95 lipid synthesis by AEA was not modified by the PKC inhibitor GF109203X (34 , 38) or by the PI3K inhibitor wortmannin (48) . In contrast, the MAPK inhibitor PD098059 (47) significantly (P<0.05) antagonized the effect of AEA, to an extent similar to that seen in the presence of the CB2 antagonist AM-630 (see Fig. 3C ). This suggests an important involvement of the MAPK pathway in endocannabinoid-induced sebocyte lipid synthesis. This concept was further supported by the finding that both AEA and 2-AG induced a marked, transient phosphorylation of MAPK Erk-1/2, which indicated activation of the MAPK pathway (Fig. 5B, C ).
Endocannabinoids upregulate expression of genes involved in the regulation of lipid synthesis
Members of the peroxisome proliferator-activated receptor (PPAR) nuclear transcription factor family are recognized as key regulators of lipid homeostasis in various cell types (50 51 52 53) . Interestingly, recent reports directly link endocannabinoid signaling to certain PPARs (54 , 55) . Because the stimulation of lipid synthesis that we had demonstrated for endocannabinoids appears to be rather similar to that reported for selected PPAR ligands in SZ95 and SEB-1 sebocytes (37 , 56 , 57) , we therefore investigated the effect of endocannabinoid treatment on the expression of PPAR isoforms in SZ95 sebocytes. As assessed by Q-PCR analysis (Table 1 ), both AEA and 2-AG significantly (P<0.05) up-regulated the expression of PPARδ and PPARγ, whereas PPARα gene expression was only increased in the 2-AG-treated group, at the 24-h time point.
Furthermore, both in human sebaceous glands and in cultured human SZ95 sebocytes, we have recently described (unpublished results) the expression pattern of recognized target genes (such as ADRP and PGAR), which are regulated or induced by PPARγ in macrophages, adipocytes, or dendritic cells (58 59 60). In addition, COX2 was also defined as PPARγ-induced target gene in SZ95 sebocytes (61) . Hence, we finally also tested whether endocannabinoid treatment affects the expression of selected target genes. Intriguingly, both AEA and 2-AG dramatically elevated (i.e., 10- to 15-fold increase at 24 h) the expression levels of all target genes investigated, further supporting the activation of PPARγ (Table 1) . This idea was further strengthened by the observation that the effect of AEA to stimulate lipid accumulation in SZ95 sebocytes was significantly prevented by GW9662, a selective inhibitor of PPARγ (61) (Fig. 5A ).
DISCUSSION
In an effort to explore the functional significance of the ECS in human skin physiology, in the current study, we have focused on the sebaceous compartment of its adnexal structures–an emerging, major neuroendocrine organ (19 , 20 , 23 , 26) . In this context, we provide here the first evidence that prototypic endocannabinoids (AEA, 2-AG) are produced by sebocytes, and show that these endocannabinoids (at physiologically relevant concentration) stimulate sebocyte lipid synthesis and apoptosis in a CB2-mediated manner. We also provide evidence suggesting that intrasebocyte signaling downstream of CB2 stimulation involves the MAPK pathway and various nuclear transcription factors well recognized in the regulation of lipid synthesis. Taken together, our data support the concept that human sebocytes are both sources and targets of endocannabinoids, where they function as constitutively active paracrine-autocrine positive regulators of sebaceous gland lipid homeostasis and negative regulators of sebocyte survival.
Previous reports have shown that endocannabinoids may exert their cellular actions via CB1, CB2, and, in the case of AEA, TRPV1 receptors (3 , 18 , 44 , 45) . Furthermore, we have recently described the functional existence of TRPV1 both in human sebaceous glands and in SZ95 sebocytes (36) . Therefore, an important goal of our study was to identify the molecular targets of the endocannabinoids. Several lines of evidence indicate the endocannabinoids induce lipid synthesis and cell death in sebaceous cells exclusively via CB2 receptors. First, the cellular effects of endocannabinoids were abrogated by the CB2-specific antagonist AM-630 but not by TRPV1 or CB1 antagonists. Second, endocannabinoids were ineffective in sebocytes in which CB2 expression was selectively knocked down by RNAi. Third, the effects of endocannabinoids were mimicked by the synthetic CB2-specific agonist JWH-015 but not by the CB1-specific agonist ACEA. Fourth, CB2 was successfully identified in SZ95 sebocytes (both at the protein and mRNA levels), whereas the expression of CB1 was uncertain.
We also intended to define the downstream signaling mechanisms activated by CB2. Among the multiple intracellular signal transduction systems (e.g., PKC, MAPK, PI3K) known to be modulated by cannabinoids (47 48 49) , we identified the MAPK pathway as a mediator of CB2-induced cellular actions of endocannabinoids in human sebocytes. We also show for the first time that CB2 activation in sebocytes also results in the induction PPAR isoforms and certain target genes (of PPARγ) involved in regulating lipid homeostasis in various cell types (50 51 52 53) . This suggests that the ECS may modulate the lipogenic gene expression profile of the cells.
Our preclinical data in one of the best established human sebocyte cell culture systems encourage the systematic exploration now of whether CB2 antagonists or agonists can be exploited in the management of common skin disorders that are characterized by sebaceous gland dysfunctions (e.g., acne vulgaris, seborrhea, dry skin, sebaceous gland-derived tumors). The observed enhancement of lipid synthesis induced by endocannabinoids strikingly resemble those seen in acne vulgaris, a common, multifactorial pilosebaceous inflammatory skin disease in which differentiation and hence lipid synthesis of sebocytes are pathologically increased (21 , 23 , 24) . This, and the finding in CB2-silenced SZ95 sebocytes that both AEA-stimulated and basal lipid synthesis, were suppressed may be interpreted to indicate a role of enhanced CB2 signaling in acne pathogenesis. Proof-of-principle studies are now warranted to test the therapeutic value of CB2 blockade in the clinical management e.g., of acne and seborrhea. Conversely, CB2 agonists deserve exploration as novel therapeutic tools for enhancing sebum production in excessively dry skin and/or for stimulating sebocyte apoptosis in sebaceous tumors.
CB1-mediated signaling inhibits human hair growth and induces apoptosis-driven regression in the hair follicle (18) . Furthermore, it also suppresses differentiation of epidermal keratinocytes (10 , 13) . Moreover, both CB1 and CB2-coupled mechanisms have been reported to suppress murine (15) and human (16) skin tumor growth and to attenuate murine allergic contact dermatitis (17) . Our current data suggest that the endogenously active ECS, which enhances sebocyte lipid synthesis and cell death selectively operates via CB2. This, in turn, suggests the existence of cell type-specific and receptor-selective regulatory endocannabinoid mechanisms in mammalian skin, which call for systematic further experimental dissection. Moreover, in view of growing insights into the importance of neuroendocrine cross-talks, e.g., between cannabinoids/CBs and melanocortin receptors (63) , and of the multiple neuroendocrine controls that human sebocytes are subject to (23 , 24 , 26) , including melanocortin receptor-mediated ones (63) , it also deserves to be dissected whether and how CB2-mediated signaling is regulated by other sebaceous neuroendocrine signals and/or modulates their production/activity.
In summary, we have shown the expression of functional CB2 receptors and of key endocannabinoids by human sebocytes, which may utilize this endogenous cannabinoid signaling system for the autocrine and paracrine control of sebocyte lipid production and death in a MAPK pathway-dependent manner.
Source, Graphs and Figures: Endocannabinoids enhance lipid synthesis and apoptosis of human sebocytes via cannabinoid receptor-2-mediated signaling
We had previously shown that both locally produced endocannabinoids and exocannabinoids, via cannabinoid receptor-1 (CB1), are powerful inhibitors of human hair growth. To further investigate the role of the cannabinoid system in pilosebaceous unit biology, we have explored in the current study whether and how endocannabinoids have an impact on human sebaceous gland biology, using human SZ95 sebocytes as cell culture model. Here, we provide the first evidence that SZ95 sebocytes express CB2 but not CB1. Also, prototypic endocannabinoids (arachidonoyl ethanolamide/anandamide, 2-arachidonoyl glycerol) are present in SZ95 sebocytes and dose-dependently induce lipid production and (chiefly apoptosis-driven) cell death. Endocannabinoids also up-regulate the expression of key genes involved in lipid synthesis (e.g., PPAR transcription factors and some of their target genes). These actions are selectively mediated by CB2-coupled signaling involving the MAPK pathway, as revealed by specific agonists/antagonists and by RNA interference. Because cells with "silenced" CB2 exhibited significantly suppressed basal lipid production, our results collectively suggest that human sebocytes utilize a paracrine-autocrine, endogenously active, CB2-mediated endocannabinoid signaling system for positively regulating lipid production and cell death. CB2 antagonists or agonists therefore deserve to be explored in the management of skin disorders characterized by sebaceous gland dysfunctions (e.g., acne vulgaris, seborrhea, dry skin).–Dobrosi, N., Tóth, B. I., Nagy, G., Dózsa, A., Géczy, T., Nagy, L., Zouboulis, C. C., Paus, R., Kovács, L., BÃró, T. Endocannabinoids enhance lipid synthesis and apoptosis of human sebocytes via cannabinoid receptor-2-mediated signaling.
INTRODUCTION
THE ENDOCANNABINOID SYSTEM (ECS), that is, endocannabinoids (such as arachidonoyl ethanolamide, AEA, and 2-arachidonoyl glycerol, 2-AG), specific cannabinoid receptors (CB1 and CB2), and enzymes involved in the synthesis and degradation of endocannabinoids, has emerged as a versatile modulatory system, implicated in a plethora of physiological and pathophysiological regulatory mechanisms (12 3) . Classically, CB1-mediated effects were mostly described in the central nervous system and were shown to involve regulating e.g., synaptic functions, memory, and motor learning (3 4 5 6) . Peripherally, the ECS has become implicated in regulation of e.g., immune and cardiovascular processes, apparently chiefly via CB2-coupled signaling (7 8 9) .
Recent intriguing findings, however, have also identified the functional existence of various members of the ECS on numerous other, previously unappreciated cell and tissue types. Among these, we have been particularly interested in human and murine skin and related cutaneous phenomena. Elements of the ECS were extensively documented in epidermal keratinocytes (10 11 12) . Moreover, cannabinoids were shown to suppress in vitro proliferation (and differentiation) of cultured epidermal keratinocytes (11 , 13) , similar to the effects of selective CB2 agonists on human coronary artery smooth muscle cells (14) , as well as the in vivo growth of murine skin tumors (15) and human melanomas (16) . In addition, using double CB1/CB2 gene-deficient mice, Karsak et al. (17) have elegantly demonstrated that endocannabinoids attenuate allergic contact dermatitis.
We have recently identified a CB1-mediated mechanism for nonclassical, peripheral tissue activities of endocannabinoids and exocannabinoids in the human system (18) . Namely, using organ-cultured human scalp hair follicles, we have shown that locally produced endocannabinoids (via CB1 that is expressed mainly on outer root sheath keratinocytes of the follicle) inhibit human hair growth and induce premature apoptosis-driven involution of this complex miniorgan (catagen).
Hair follicles most commonly are arranged in pilosebaceous units, which display another characteristic adnexal structure of mammalian skin, i.e., the sebaceous gland (reviewed in refs. 19 , 20 ). It is extensively documented that epithelial cells of the sebaceous gland, i.e., sebocytes, play a central role in the regulation of cutaneous lipid homeostasis and that pathological malfunctions of these cells may result in such common cutaneous diseases as acne vulgaris (21 22 2324) .
Of importance, sebaceous gland cells reportedly also show CB receptor immunoreactivity in situ (12) . However, apart from anecdotes about marijuana users often developing acne, direct evidence for the presence of functional ECS in sebaceous glands and a description of the potential effects of endocannabinoids on various biological processes of human sebocytes are lacking.
In the current study, we have therefore analyzed the presence and function of the ECS and related signaling mechanisms in human sebaceous gland-derived cells, using the SZ95 sebocyte cell line, one of the best-established human sebocyte cell culture models (25 26 27) . Specifically, we intended to clarify which CB receptors are expressed by human sebocytes, and whether prototypic endocannabinoids (AEA, 2-AG) can be detected in SZ95 sebocytes. Furthermore, we have evaluated the effects of endocannabinoids on defined sebocyte functions (e.g., lipid synthesis, cell growth and death, gene expression), using an array of cellular and molecular assays, and have defined the involvement of various intracellular signaling pathways in mediating the effects of endocannabinoids (by employing specific agonists, antagonists, and RNA interference).
Collectively, these studies provide the first evidence that human sebocytes selectively express functional CB2, contain key endocannabinoids, and utilize this endogenous cannabinoid signaling system for the autocrine and paracrine control of human sebocyte lipid production and death.
MATERIALS AND METHODS
Materials
AEA, 2-AG, arachidonyl-2-chloroethylamide (ACEA), AM-251, iodo-resiniferatoxin (I-RTX), and GF10203X were purchased from Sigma-Aldrich (Taufkirchen, Germany). JWH-015 and GW9662 were obtained from Cayman Chemical Company (Ann Arbor, MI, USA). PD098059 and wortmannin were obtained from Calbiochem (Nottingham, UK), and AM630 was from Tocris Bioscience (Ellisville, MO, USA).
Cell culturing
Human immortalized SZ95 sebocytes (25 26 27) , were cultured in Sebomed® basal medium (Biochrom, Berlin, Germany) supplemented with 10% fetal bovine serum (Invitrogen, Paisley, UK), 1 mM CaCl2, 5 ng/ml human recombinant epidermal growth factor (Sigma-Aldrich), 50 U/ml penicillin, and 50 µg/ml streptomycin (both from Biogal, Debrecen, Hungary).
Determination of endocannabinoid levels
The levels of endocannabinoids (AEA, 2-AG) were determined by liquid chromatography/in line mass spectrometry as described in our earlier reports (18 , 28) .
Immunocytochemistry
SZ95 sebocytes were fixed in acetone, permeabilized by 0.1% Triton-X-100 (Sigma-Aldrich) in phosphate-buffered saline (PBS), and then incubated with rabbit anti-CB1 or anti-CB2 primary antibodies (Cayman) for 60 min (dilution 1:200). Slides were then incubated with a goat anti-rabbit fluorescein-isothiocyanate (FITC) -conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA) (dilution 1:200), and nuclei were visualized using DAPI (Vector). Cells were examined on a Nikon Eclipse E600 fluorescent microscope (Nikon, Tokyo, Japan) (29) .
Immunohistochemistry
The study was approved by the Institutional Research Ethics Committee and adhered to Declaration of Helsinki guidelines. Normal female trunk skin samples, obtained during plastic surgery, were fixed in 4% formalin and embedded in paraffin, and 4-µm-thick sections were obtained. After antigen retrieval and blocking of the endogenous peroxidase activity, tissue sections were incubated with the above CB1 and CB2 primary antibodies (Cayman) (dilution 1:150 for CB1, 1:200 for CB2). Sections were then incubated with the EnVision+ System Labeled polymer-HRP Anti-Rabbit (Dako, Glostrup, Denmark) with 3,3'-diaminobenzidine (DAB) visualization techniques. Tissue samples were finally counterstained with hematoxylin (Sigma-Aldrich) and mounted in aqueous mounting medium (Dako) (18 , 29 , 30) .
In the course of the immunohistochemistry, numerous control experiments were performed. As "staining-negative" controls, the appropriate primary antibodies were either omitted from the procedure or were preabsorbed with synthetic blocking peptides (Cayman) (see Fig. 1 ). As "tissue-negative" controls, CB-labeling was performed on tissues not expressing CB1 (human mast cell line HMC-1) (31) or CB2 (HMC-1, human hair follicle) (18 , 31) . In addition, the specificity of CB receptor staining was also measured 1) on tissues recognized to be CB1 (HaCaT epidermal keratinocytes, hair follicle) (10 , 13 , 18) or CB2 (HaCaT keratinocytes, human peripheral monocytes, spleen) positive (11 , 18 , 32) (data not shown); or 2) by employing another set of antibodies against CB1 and CB2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The application of these latter primary antibodies resulted in identical staining patterns (data not shown).
Western blot analysis
Western immunoblotting was performed as described in our earlier reports (29 , 30 , 33) . In brief, cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to BioBond nitrocellulose membranes (Whatman, Maidstone, England), and then probed with anti-CB1 or anti-CB2 receptor antibodies (1:200, Cayman); with a rabbit antibody against the mitogen-activated protein kinase (MAPK) Erk-1/2 (Santa Cruz); or a mouse antibody recognizing the phosphorylated form of Erk-1/2 (pErk-1/2, Santa Cruz) (1:1500 dilution in both cases). Horseradish peroxidase—polymer-conjugated, respective anti-rabbit or anti-mouse IgG antibodies (Envision labeling, DAKO) were used as secondary antibodies, and the immunoreactive bands were visualized by SuperSignal West Pico Chemiluminescent Substrate-enhanced chemiluminescence (Pierce, Rockford, IL, USA). Immunoblots were then subjected to densitometric analysis using an Intelligent Dark Box (Fuji, Tokyo, Japan) and the Image Pro Plus 4.5.0 software (Media Cybernetics, Silver Spring, MD, USA). To assess equal loading, membranes were stripped and then reprobed with a rabbit cytochrome-C (Cyt-C) antibody (Santa Cruz) followed by a similar visualization procedure as described above.
Reverse transcriptase-polymerase chain reaction (RT-PCR)
The expression of CB1 and CB2 receptor mRNA was determined by semiquantitative RT-PCR, as we have described before (18 , 29 , 30) . In brief, isolated total RNA was reverse-transcribed into cDNA and then amplified on a GeneAmp PCR System 2400 DNA Thermal Cycler (Applied Biosystems, Foster City, CA, USA) using optimized thermal protocols. Primers were synthesized by Invitrogen (CB1, forward: CAAGCCCGCATGGACATTAGGTTA, CB1, reverse: TCCGAGTCCCCCATGCTGTTATC; CB2, forward: TCCCACTGATCCCCAATGACTACC, CB2, reverse: AGGATCTCGGGGCTTCTTCTTTTG; glyceraldehyde 3-phosphate dehydrogenase (GAPDH), forward: ATGGTGAAGGTCGGTGTGAAC, GAPDH, reverse: GCTGACAATCTTGAGGGAGT). PCR products were visualized on 1.5% agarose gel with ethidium bromide (0.5 mg/ml, Sigma-Aldrich) under UV, and the photographed bands were quantified by Image Pro Plus 4.5.0 software.
Quantitative real-time PCR
Quantitative PCR (Q-PCR) was performed on an ABI Prism 7000 sequence detection system (Applied Biosystems) using the 5' nuclease assay, as detailed in our previous report (18 , 29) . To detect the expression of genes involved in the regulation of lipid synthesis (see also Table 1 ), the following TaqMan primers and probes were used: peroxisome proliferator-activated receptor (PPAR)α, forward: CATTACGGAGTCCACGCGT, PPARα, reverse: ACCAGCTTGAGTCGAATCGTT; PPARα, probe FAM-AGGCTGCAAGGGCTTCTTTCGGCG-TAMRA; PPARδ, forward: AGCATCCTCACCGGCAAAG, PPARδ, reverse: CCACAATGTCTCGATGTCGTG; PPARδ, probe FAM-CAGCCACACGGCGCCCTTTG-TAMRA; PPARγ, forward: GATGACAGCGACTTGGCAA, PPARγ, reverse: CTTCAATGGGCTTCACATTCA; PPARγ, probe FAM-CAAACCTGGGCGGTCTCCACTGAG-TAMRA; adipose differentiation-related protein (ADRP), forward: TGACTGGCAGTGTGGAGAAGA, ADRP, reverse: ATCATCCGACTCCCCAAGA; ADRP, probe FAM-TCTGTGGTCAGTGGCAGCATTAACACA-TAMRA; PPARγ-regulated angiopoietin-related protein (PGAR), forward: TCCGCAGGGACAAGAACTG, PGAR, reverse: CGGAAGTACTGGCCGTTGA; PGAR, probe FAM-TTGGAATGGCTGCAGGTGCCA-TAMRA; and a predesigned assay for cyclooxygenase-2 (COX-2) (Applied Biosystems, assay ID: Hs00153133_m1). As internal controls, transcripts of human cyclophilin-A (forward: ACGGCGAGCCCTTGG, reverse: TTTCTGCTGTCTTTGGGACCT; probe FAM-CGCGTCTCCTTTGAGCTGTTTGCA-TAMRA) were determined.
RNA interference (RNAi)
SZ95 sebocytes were seeded in 6-well culture plates in SeboMed medium lacking antibiotics. At 50—70% confluence, medium was replaced by serum-free OptiMEM (Invitrogen), and cells were transfected with various CB2-specific Stealth RNAi oligonucleotides (ID: HSS102085, HSS102086, HSS102087, Invitrogen) (40 nM) using Lipofectamine 2000 transfection reagent (Invitrogen). For controls, RNAi Negative Control Duplexes (Scrambled RNAi, Invitrogen) and CB1-specific Stealth RNAi oligonucleotides (ID: HSS102082) were employed. Three hours after transfection, medium was replaced by complete Sebomed® medium, and cells were allowed to recover for 24 h. The efficacy of siRNA-driven knockdown was daily evaluated by RT-PCR and Western blot analysis for 4 days (34) .
Determination of intracellular lipids
For semiquantitative detection of sebaceous lipids, SZ95 sebocytes were cultured on glass coverslips and treated with several compounds (AEA, 2-AG) for 24—48 h. Cells were fixed in 4% paraformaldehyde (Sigma-Aldrich), washed in 60% isopropanol, and stained in Oil Red O solution (0.3% in isopropanol) (Sigma-Aldrich). Cells were counterstained with hematoxylin (Sigma-Aldrich) and were mounted in aqueous mounting medium (Dako) (35 , 36) .
For quantitative measurement, SZ95 sebocytes (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One, Frickenhausen, Germany) in quadruplicates and were treated with compounds for 24—48 h. Supernatants were then discarded, and 100 µl of 1 µg/ml Nile red (Sigma-Aldrich) solution was added to each well. The emitted fluorescence was measured on a Molecular Devices FlexStation 384II Fluorescence Image Microplate Reader (FLIPR, Molecular Devices, San Francisco, CA, USA). Results are presented as percentages of the relative fluorescence units (RFU) in comparison with the controls, using 485-nm excitation and 565-nm emission wavelengths for neutral lipids, and 540-nm excitation and 620-nm emission wavelengths for polar lipids (35 3637) .
Determination of viable cell number
The number of viable cells was determined by measuring the conversion of the tetrazolium salt MTT (Sigma-Aldrich) to formazan by mitochondrial dehydrogenases. Cells were plated in 96-well plates (15,000 cells/well) in quadruplicates and were cultured for 24—48 h. Cells were then incubated with 0.5 mg/ml MTT, and the concentration of formazan crystals (as an indicator of the viable cell number) was determined colorimetrically (A550), according to our previous reports (29 , 33 , 34 , 38) .
Determination of apoptosis
Abolishment of mitochondrial membrane potential is one of the earliest markers of apoptosis (39) . To assess the process, mitochondrial membrane potential of SZ95 sebocytes was determined using a MitoProbe DilC1(5) Assay Kit (Invitrogen). Cells (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) in quadruplicate and were treated with various compounds for 24—48 h. After removal of supernatants, cells were incubated with DilC1(5) working solution (30 µl/well), and the fluorescence of DilC1(5) was measured at 630-nm excitation and 670-nm emission wavelengths using FLIPR (36) .
In addition, another hallmark of apoptosis (i.e., membrane perturbation) was also assessed by flow cytometry according to our previous reports (29 , 34 , 36 , 38) . In brief, following treatment with various agents, SZ95 sebocytes were harvested and stained with an Annexin-V-FITC/propidium iodide (PI) apoptosis kit (Sigma-Aldrich) following the manufacturer's protocol. Fluorescence intensity was measured by a Coulter Epics XL (Beckman Coulter, Fullerton, CA, USA) flow cytometer.
Determination of necrosis/cytotoxicity
Necrotic cell death was first determined by measuring the glucose-6-phosphate-dehydrogenase (G6PD) release (G6PD Release Assay Kit, Invitrogen). The enzyme activity was detected by a 2-step enzymatic process that leads to the reduction of resazurin into red-fluorescent resorufin. SZ95 sebocytes (15,000 cells/well) were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) in quadruplicate and treated with various compounds for 24—48 h. A 2x reaction medium was then prepared according to the manufacturer's protocol and was added to the wells in 1:1 dilution. The fluorescence emission of resorufin was monitored by FLIPR at 545-nm excitation and 590-nm emission wavelengths (36) .
The cytotoxic effects of cannabinoid treatment were also determined by Sytox Green staining (Invitrogen). The dye is able to penetrate (and then bind to the nucleic acids) only to necrotic cells with ruptured plasma membranes, whereas healthy cells with intact surface membranes show negligible Sytox Green staining. SZ95 sebocytes were cultured in 96-well black-well/clear-bottom plates (Greiner Bio One) and were treated with various compounds (AEA, 2-AG, JWH-015, ACEA) for 24—48 h. Supernatants were then discarded, and the cells were incubated with 1 µM Sytox Green solution. The fluorescence of Sytox Green was measured at 490-nm excitation and 520-nm emission wavelengths using FLIPR (36) .
Statistical analysis
When applicable, data were analyzed using a 2-tailed unpaired t test, and values of P < 0.05 were regarded as significant. In addition, statistical differences were further verified using 1-way ANOVA with Bonferroni and Dunnett post hoc probes, resulting in similar results (data not shown).
RESULTS
Human sebaceous gland epithelium and human SZ95 sebocytes express CB receptors
First, we intended to identify the existence of CB receptors in human sebaceous gland in situ and on human SZ95 sebocytes. Similar to a previous report (12) , in situ, both CB1- and CB2-like immunoreactivity was identified in the sebaceous gland epithelium of normal human scalp skin sections (Fig. 1A, B ). However, using mutually complementary and confirmatory immunocytochemistry and Western blot analysis on human SZ95 sebocytes, we were able to identify only the relatively high expression of CB2, whereas the appearance of CB1-like immunoreactivity, in all cases, was around the detection limit (Fig. 1C-E ). In addition, supporting the above findings, transcription of the CB2 (but not of CB1) gene in SZ95 sebocytes was demonstrated by RT-PCR (Fig. 1F ) and by real-time Q-PCR (not shown). For positive controls, human organ-cultured hair follicles (for CB1) (18) and human peripheral monocytes (for CB2) (32) (as well as other tissue and cell types listed in Materials and Methods, data not shown) were employed. Therefore, in striking contrast to human hair follicle epithelium in situ (18) , human sebocytes express primarily, if not exclusively, CB2, at least in vitro.
Human SZ95 sebocytes produce endocannabinoids
We also tested whether SZ95 sebocytes synthesize endocannabinoids. Using mass spectrometry, we were able to show that SZ95 sebocytes express both AEA (66.7±10 fmol/mg tissue) and 2-AG (6.2±1 pmol/mg tissue) (means±SE, n=4) at levels similar to those detected earlier in various skin samples; e.g., ∼50 fmol/mg tissue AEA in rat paw and mouse ear samples (17 , 40) or 20—30 pmol/mg tissue 2-AG in mouse ear skin (17) .
Endocannabinoids enhance lipid synthesis and induce apoptosis in SZ95 sebocytes
To further explore the functionality and the biological consequences of CB stimulation as well as possible auto- and paracrine-signaling events, we next investigated the effects of these endocannabinoids found in human SZ95 sebocytes on key functions of these cells. As revealed by Oil Red-O staining and quantitative Nile Red-based fluorescence assay, both AEA and 2-AG markedly (P<0.05) and dose dependently enhanced neutral lipid accumulation in SZ95 sebocytes (Fig. 2A, B ), reflecting stimulation of sebocyte differentiation (35) .
By MTT assay, we also showed that stimulation with these endocannabinoids decreased sebocyte viability (Fig. 2C ). To assess whether this effect was due to apoptosis and/or necrosis, first, flow cytometry analysis was performed. As seen in Fig. 2D , both AEA and 2-AG markedly (P<0.05) increased the number of Annexin-V-positive cells (reflecting phosphatidyl-serine translocation) (29 , 34 , 36 , 38 , 41) , whereas the number of double Annexin-V- and PI-positive SZ95 sebocytes was only slightly elevated. This suggests that exogenously applied endocannabinoids primarily induce sebocyte apoptosis. In further support of this concept, in a series of quantitative fluorimetric assays, AEA (as well as 2-AG, data not shown) significantly decreased mitochondrial membrane potential (another hallmark of apoptosis) (36 , 39) in a dose-dependent fashion (P<0.05), while only the highest concentration of AEA was able to moderately (yet significantly, P<0.05) increase Sytox Green accumulation and G6PD release, two complementary indicators of necrosis/cytotoxicity (Fig. 2C ) (36) .
The effects of AEA on human sebocytes are selectively mediated by CB2
We then investigated whether the cellular actions of AEA on SZ95 sebocytes were mediated by the CB receptors. As seen in Fig. 3A, B , the synthetic CB2-specific agonist JWH-015 (3 , 7 , 42) –but, notably, not the CB1-specific agonist ACEA (3 , 7 , 43) –mimicked the action of endocannabinoids to enhance lipid synthesis and to induce (chiefly apoptosis-driven) cell death. Moreover, the effects of AEA were prevented by the CB2-specific antagonist AM-630 (3 , 7 , 11 , 28) (Fig. 3C, D ) but not by the CB1-specific inhibitor AM-251 (3 , 7 , 18) (Fig. 3C ). These data suggested that the sebocyte-modulatory effects of endocannabinoids are mediated by CB2 but not by CB1.
However, AEA, in various cellular systems, may also act on another receptor, i.e., transient receptor potential vanilloid-1 (TRPV1, the capsaicin receptor) (3 , 44 , 45) . Furthermore, we have recently described the functional existence of TRPV1-mediated signaling both in human organ-cultured hair follicles (29) and in SZ95 sebocytes (36) . Therefore, to further dissect the exact cellular mechanisms of action of the endocannabinoid, we also measured the effect of the TRPV1 antagonist I-RTX (46) on the actions of AEA. As seen in Fig. 3C , I-RTX did not interfere with the ability of AEA to enhance lipid synthesis (and to induce cell death, data not shown).
Collectively, these functional data are in agreement with the apparent lack of CB1 expression in SZ95 sebocytes at either the protein or gene level (Fig. 1) and suggest that the cellular actions of endocannabinoids are mediated by CB2. To further assess the role of CB2, a series of RNAi experiments against the receptors was carried out (Fig. 4A, B ). Western blot and RT-PCR analysis revealed that the expression of CB2 was significantly knocked down by all 3 RNAi probes at day 2 after transfection and remained suppressed also on day 3. However, this phenomenon was reversible, because we observed a return of the immunosignals at day 4. Scrambled RNAi probes (Fig. 4A ) or RNAi oligonucleotides against CB1 (data not shown) had no effect on the expression of CB2, indicating the specificity of the CB2 knockdown.
We then investigated the effects of AEA-treatment (24 h) on the lipid synthesis of RNAi-transfected SZ95 sebocytes on day 2. Similar to the effects of CB inhibitors, RNAi knockdown of CB2 resulted in the loss of effect of AEA in enhancing lipid synthesis. In contrast, treatment with the CB1-targeted RNAi did not affect the cellular action of AEA (Fig. 4C ). Intriguingly, in SZ95 sebocytes with RNAi-mediated knockdown of CB2, we found markedly and significantly decreased basal lipid content as well (Fig. 4C ). This strongly suggests that CB2-mediated signaling indeed plays an important endogenous role in the constitutive regulation of sebocyte lipid synthesis.
The CB2-mediated cellular signaling involves the MAPK pathway
On various cell types, CB receptor-mediated signaling, initiated by either endocannabinoids or exocannabinoids, recruits multiple intracellular pathway systems, such as protein kinase C (PKC), MAPK, or phosphatidyl-inositol-3-kinase (PI3K) (47 48 49) . Therefore, we have also undertaken attempts to elucidate selected components of CB2-mediated intracellular signaling in human sebocytes.
As seen in Fig. 5A , the enhancement of SZ95 lipid synthesis by AEA was not modified by the PKC inhibitor GF109203X (34 , 38) or by the PI3K inhibitor wortmannin (48) . In contrast, the MAPK inhibitor PD098059 (47) significantly (P<0.05) antagonized the effect of AEA, to an extent similar to that seen in the presence of the CB2 antagonist AM-630 (see Fig. 3C ). This suggests an important involvement of the MAPK pathway in endocannabinoid-induced sebocyte lipid synthesis. This concept was further supported by the finding that both AEA and 2-AG induced a marked, transient phosphorylation of MAPK Erk-1/2, which indicated activation of the MAPK pathway (Fig. 5B, C ).
Endocannabinoids upregulate expression of genes involved in the regulation of lipid synthesis
Members of the peroxisome proliferator-activated receptor (PPAR) nuclear transcription factor family are recognized as key regulators of lipid homeostasis in various cell types (50 51 52 53) . Interestingly, recent reports directly link endocannabinoid signaling to certain PPARs (54 , 55) . Because the stimulation of lipid synthesis that we had demonstrated for endocannabinoids appears to be rather similar to that reported for selected PPAR ligands in SZ95 and SEB-1 sebocytes (37 , 56 , 57) , we therefore investigated the effect of endocannabinoid treatment on the expression of PPAR isoforms in SZ95 sebocytes. As assessed by Q-PCR analysis (Table 1 ), both AEA and 2-AG significantly (P<0.05) up-regulated the expression of PPARδ and PPARγ, whereas PPARα gene expression was only increased in the 2-AG-treated group, at the 24-h time point.
Furthermore, both in human sebaceous glands and in cultured human SZ95 sebocytes, we have recently described (unpublished results) the expression pattern of recognized target genes (such as ADRP and PGAR), which are regulated or induced by PPARγ in macrophages, adipocytes, or dendritic cells (58 59 60). In addition, COX2 was also defined as PPARγ-induced target gene in SZ95 sebocytes (61) . Hence, we finally also tested whether endocannabinoid treatment affects the expression of selected target genes. Intriguingly, both AEA and 2-AG dramatically elevated (i.e., 10- to 15-fold increase at 24 h) the expression levels of all target genes investigated, further supporting the activation of PPARγ (Table 1) . This idea was further strengthened by the observation that the effect of AEA to stimulate lipid accumulation in SZ95 sebocytes was significantly prevented by GW9662, a selective inhibitor of PPARγ (61) (Fig. 5A ).
DISCUSSION
In an effort to explore the functional significance of the ECS in human skin physiology, in the current study, we have focused on the sebaceous compartment of its adnexal structures–an emerging, major neuroendocrine organ (19 , 20 , 23 , 26) . In this context, we provide here the first evidence that prototypic endocannabinoids (AEA, 2-AG) are produced by sebocytes, and show that these endocannabinoids (at physiologically relevant concentration) stimulate sebocyte lipid synthesis and apoptosis in a CB2-mediated manner. We also provide evidence suggesting that intrasebocyte signaling downstream of CB2 stimulation involves the MAPK pathway and various nuclear transcription factors well recognized in the regulation of lipid synthesis. Taken together, our data support the concept that human sebocytes are both sources and targets of endocannabinoids, where they function as constitutively active paracrine-autocrine positive regulators of sebaceous gland lipid homeostasis and negative regulators of sebocyte survival.
Previous reports have shown that endocannabinoids may exert their cellular actions via CB1, CB2, and, in the case of AEA, TRPV1 receptors (3 , 18 , 44 , 45) . Furthermore, we have recently described the functional existence of TRPV1 both in human sebaceous glands and in SZ95 sebocytes (36) . Therefore, an important goal of our study was to identify the molecular targets of the endocannabinoids. Several lines of evidence indicate the endocannabinoids induce lipid synthesis and cell death in sebaceous cells exclusively via CB2 receptors. First, the cellular effects of endocannabinoids were abrogated by the CB2-specific antagonist AM-630 but not by TRPV1 or CB1 antagonists. Second, endocannabinoids were ineffective in sebocytes in which CB2 expression was selectively knocked down by RNAi. Third, the effects of endocannabinoids were mimicked by the synthetic CB2-specific agonist JWH-015 but not by the CB1-specific agonist ACEA. Fourth, CB2 was successfully identified in SZ95 sebocytes (both at the protein and mRNA levels), whereas the expression of CB1 was uncertain.
We also intended to define the downstream signaling mechanisms activated by CB2. Among the multiple intracellular signal transduction systems (e.g., PKC, MAPK, PI3K) known to be modulated by cannabinoids (47 48 49) , we identified the MAPK pathway as a mediator of CB2-induced cellular actions of endocannabinoids in human sebocytes. We also show for the first time that CB2 activation in sebocytes also results in the induction PPAR isoforms and certain target genes (of PPARγ) involved in regulating lipid homeostasis in various cell types (50 51 52 53) . This suggests that the ECS may modulate the lipogenic gene expression profile of the cells.
Our preclinical data in one of the best established human sebocyte cell culture systems encourage the systematic exploration now of whether CB2 antagonists or agonists can be exploited in the management of common skin disorders that are characterized by sebaceous gland dysfunctions (e.g., acne vulgaris, seborrhea, dry skin, sebaceous gland-derived tumors). The observed enhancement of lipid synthesis induced by endocannabinoids strikingly resemble those seen in acne vulgaris, a common, multifactorial pilosebaceous inflammatory skin disease in which differentiation and hence lipid synthesis of sebocytes are pathologically increased (21 , 23 , 24) . This, and the finding in CB2-silenced SZ95 sebocytes that both AEA-stimulated and basal lipid synthesis, were suppressed may be interpreted to indicate a role of enhanced CB2 signaling in acne pathogenesis. Proof-of-principle studies are now warranted to test the therapeutic value of CB2 blockade in the clinical management e.g., of acne and seborrhea. Conversely, CB2 agonists deserve exploration as novel therapeutic tools for enhancing sebum production in excessively dry skin and/or for stimulating sebocyte apoptosis in sebaceous tumors.
CB1-mediated signaling inhibits human hair growth and induces apoptosis-driven regression in the hair follicle (18) . Furthermore, it also suppresses differentiation of epidermal keratinocytes (10 , 13) . Moreover, both CB1 and CB2-coupled mechanisms have been reported to suppress murine (15) and human (16) skin tumor growth and to attenuate murine allergic contact dermatitis (17) . Our current data suggest that the endogenously active ECS, which enhances sebocyte lipid synthesis and cell death selectively operates via CB2. This, in turn, suggests the existence of cell type-specific and receptor-selective regulatory endocannabinoid mechanisms in mammalian skin, which call for systematic further experimental dissection. Moreover, in view of growing insights into the importance of neuroendocrine cross-talks, e.g., between cannabinoids/CBs and melanocortin receptors (63) , and of the multiple neuroendocrine controls that human sebocytes are subject to (23 , 24 , 26) , including melanocortin receptor-mediated ones (63) , it also deserves to be dissected whether and how CB2-mediated signaling is regulated by other sebaceous neuroendocrine signals and/or modulates their production/activity.
In summary, we have shown the expression of functional CB2 receptors and of key endocannabinoids by human sebocytes, which may utilize this endogenous cannabinoid signaling system for the autocrine and paracrine control of sebocyte lipid production and death in a MAPK pathway-dependent manner.
Source, Graphs and Figures: Endocannabinoids enhance lipid synthesis and apoptosis of human sebocytes via cannabinoid receptor-2-mediated signaling
