Molecular Vision 2007; 13:1641-1650 <http://www.molvis.org/molvis/v13/a183/> Received 18 July 2007 | Accepted 10 September 2007 | Published 11 September 2007

Download Reprint

Molecular and metabolic retinoid pathways in the human ocular surface

Hachemi Nezzar,^1,2 Frédéric Chiambaretta,^1,2 Geoffroy Marceau,^1,3 Loïc Blanchon,³ Babacar Faye,³ Pierre Dechelotte,⁴ Danièle Rigal,^1,2 Vincent Sapin^1,3
 
 

¹Université d'Auvergne, JE 2447, Ardemo; ²CHU Clermont-Ferrand, Ophtalmologie; ³INSERM, U.384, Laboratoire de Biochimie, Faculté de Médecine; ⁴CHU Clermont-Ferrand, Service d'Anatomie et de Cytologie Pathologiques, Hôtel-Dieu, Clermont-Ferrand, France

Correspondence to: Pr Vincent Sapin, INSERM U.384, Laboratoire de Biochimie Médicale, Faculté de Médecine, 28 Place Henri Dunant, BP. 38, 63000 Clermont-Ferrand, France; Phone: + 00 33 4 73 17 81 74; FAX: + 00 33 4 73 27 61 32; email: vincent.sapin@u-clermont1.fr

Abstract

Purpose: To maintain its integrity, the human ocular surface (cornea and conjunctiva) has an absolute requirement for vitamin A and its active derivatives, the retinoic acids. These retinoids regulate transcriptional levels of target genes through the activation of members of a super-family of ligand-dependant nuclear receptors that feature retinoic acid receptors (RAR) α, β, and γ as well as retinoid X receptors (RXR) α, β, and γ. The expression patterns of these receptors have been partial characterized in rabbit, mouse, and human cornea and conjunctiva, but systematic tissue and cellular expression of the three RARs and three RXRs had to be completed at the adult human ocular surface. The first objective of our work was to define their expression patterns in term of genes and proteins for total human conjunctiva, cornea, and the major cell types comprising the adult human ocular surface. The second objective was to demonstrate the presence of different enzymes transforming vitamin A to retinoic acid and the functionality of this metabolic pathway in the corneal epithelium.

Methods: Total mRNA was extracted from human total cornea, conjunctiva, corneal epithelial cells (primary culture and established cell line), corneal keratocytes (primary culture), corneal endothelial cells (established cell line), and conjunctival epithelial cells (established cell line) and was submitted to reverse transcription-polymerase chain reaction (RT-PCR) analysis to determine the expression patterns of the RARs and RXRs using specific primers. Immunological staining (via histochemistry and cellular chemistry) experiments were performed to better localize RAR and RXR proteins in the ocular surface at tissue and cellular levels. We also checked mRNA expression of cellular retinol binding proteins (CRBPs) and cellular retinoic acid binding proteins (CRABPs) with the enzymes involved in retinoic acid generation, i.e., alcohol dehydrogenases (ADHs) and retinal dehydrogenases (RALDHs) or degradation (Cyp26 family members). The enzymatic generation of functional retinoids was confirmed using epithelial corneal cells treated with specific inhibitors of retinol metabolism.

Results: RAR α, RAR γ, and RXR α are expressed in the cornea, conjunctiva, and all of their constitutive cells, whereas RXR γ and RXR β were never detected in the cornea or conjunctiva. RAR β was absent in primary cultures of corneal keratinocytes. ADH3, ADH4, dehydrogenase/reductase (SDR family) 4 (DHRS4), dehydrogenase/reductase (SDR family) 9 (DHRS9), RALDH1, and RALDH3 are expressed in the ocular surface, as were the retinoid-binding proteins CRBP1, CRABP1, and CRABP2. Retinol dehydrogenase 4 (RODH4) was only detected in the conjunctiva. Corneal epithelial cells convert retinol into retinoic acid using an enzymatic pathway.

Conclusions: For the first time, we have established an exhaustive description of the expressions patterns of RARs, RXRs, ADHs, RALDHs, CRBP, and CRABPs in the human ocular surface. Our results for the human ocular surface demonstrated the presence of all the metabolic and molecular actors of the retinoic acid signaling pathway. We also demonstrated the enzymatic conversion of retinol into active retinoids in the corneal environment.

Introduction

It is well established that the cornea and the conjunctiva have an absolute requirement for vitamin A (retinol), which acts through its metabolites such as retinaldehyde, which forms the visual chromophore, and retinoic acids, which regulate gene expression. Early studies dealing with the effects of dietary vitamin A deficiency (VAD) or mouse gene targeting strategies on embryonic ocular development and malformations have demonstrated that vitamin A is required for eye morphogenesis [1,2]. Lack of vitamin A causes abnormal differentiation of the ocular surface that results in keratinization, ulceration, epithelial squamous metaplasia, and a deficiency of conjunctival goblet cells [3,4]. When xerophthalmia and keratomalacia due to vitamin A deficiency was recognized as an important public health problem in many developing countries, it triggered many studies on the use of retinoids on the ocular surface [5]. Due to these implications, topical retinoids have been tested as pharmacological agents for the treatment of ocular surface diseases involving wound healing [6], squamous metaplasia [7] or dry eye [8].

These pleiotropic effects of retinoids are mediated by retinoic acid receptors (RARs) and retinoid X receptors (RXRs), which are ligand-activated transcription factors. Two families of nuclear receptors for retinoic acid (RA) have been characterized. Members of the RAR family (α, β, and γ) are activated by most of the physiologically-occurring retinoids (all-trans RA, 9-cis RA, 4-oxo RA, and 3,4di-hydro RA). In contrast, members of the RXR family (α, β, and γ) are only activated by 9-cis RA [9,10]. The transcriptional regulation of the target genes relies on the recognition and cooperative assembly of RAR-RXR heterodimers in short DNA sequences in their promoter regions that are known as retinoic acid response elements (RARE). Several corneal and conjunctival target genes of retinoids have already been described, including proteases and their inhibitors [11], epidermal growth factor (EGF) [12], mucine [13] or AP-2 transcription factor [14]. To better understand the transcriptional regulating properties involved, partial characterization of the RARs and RXRs have been realized in rabbit, mouse, and human conjunctiva and cornea [2,15,16], but the cellular and systematic tissue expressions of these receptors had to be completed at the human ocular surface.

To be biologically active (see Figure 1), retinol must first be oxidized to retinaldehyde then to retinoic acid. A large number of enzymes catalyze the oxidation of retinol to retinaldehyde including alcohol dehydrogenases (ADH) and retinol dehydrogenase of the microsomal fraction (DHRS) as well as several enzymes able to catalyze the oxidation of retinaldehyde to retinoic acid, i.e., retinal dehydrogenase (RALDH 1, 2, 3, and 4) [17]. Retinoic acid catabolism is also an important mechanism for controlling RA levels in cells and tissues, and it is performed by three specific cytochrome P450s: CYP26A, CYP26B, and CYP26C [18]. Since retinol, retinaldehyde, and retinoic acid are lipids, they lack appreciable water solubility and consequently must be bound to proteins within cells. Several intracellular binding proteins for retinol, retinaldehyde, and retinoic acid have been identified and extensively characterized. They include cellular retinol-binding proteins type 1 and 2 (CRBP 1 and 2) and cellular retinoic acid-binding proteins type 1 and 2 (CRABP 1 and 2). Fragmental data have established the presence of retinoid-binding capacities [19] and some global retinoid metabolic pathways in the mammalian ocular surface [20]. It has also been established that tears contain detectable levels of certain retinoids [21]. Curiously, the human ocular surface metabolism of retinoids and, more particularly, the ability of human corneal epithelial cells to metabolize the retinol has not yet been documented.

The aim of this work was (1) to identify the molecular and metabolic actors involved in the retinoid pathway in the human ocular surface and (2) to evaluate retinol conversion into retinoic acid in the human corneal epithelial cells.

Methods

Collection of human tissues

Pieces of normal human bulbar conjunctiva were removed from 10 patients during ocular surgery (retina detachment, strabismus, ectropion or entropion) once informed consent was obtained, according to the tenets of the Declaration of Helsinki. The research was approved by the institutional human experimentation committee (regional ethics committee, CCP Sud-Est VI). Ten normal human corneas unsuitable for graft for serological reasons were provided by the regional corneal graft bank. All tissues appeared normal after classical histological examination. They were rinsed using cold sterile phosphate buffer saline (PBS), embedded with Tissue-Teck OCT (Sakura, Zoeterwounde, Netherlands) or put in trizol (Invitrogen, Cergy-Pontoise, France), and stored at -80 °C until use. For immunohistological experiments, 10 μm-thick cryosections were cut and mounted on Super Frost slides (Fischer Scientific, Pittsburgh, PA).

Cell cultures

A human corneal epithelium (HCE) cell line transformed with SV 40 (ATCC/CRL11135) was cultured under standard conditions (5% CO₂, 95% humidified air, 37 °C) in Dulbecco's Modified Eagle's Medium plus Ham's nutrient mixture F-12 (DMEM-F12) supplemented with 5% fetal calf serum, 5 μg/ml of insulin, 0.1 μg/ml of cholera toxin, 50 mg/ml of streptomycin, 50 IU/ml of penicillin, 0.5 mg/ml of epithelium growth factor, and 0.5% DMSO.

Primary cell cultures of human corneal epithelium were established from donor corneas obtained from the local eye bank. Briefly, the limbus was dissected from five human corneas and cut into small pieces. Explants were transferred to 25 ml culture flasks, left for 2 h to allow the epithelial cells to adhere to the support, and cultured as previously described for HCE.

Human corneal keratocyte cell cultures were established from five donor corneas. Briefly, an 8 mm diameter corneal button was punched out and the Descemet's membrane with attached endothelium was stripped off. The epithelium was also totally removed by alcohol treatment and scraping. The naked residual stroma was then cut into small pieces and treated with 0.2% collagenase B (Roche Diagnostics, Meylan, France) for 3 h at 37 °C. After centrifugation, the pellet of keratocytes was resuspended in DMEM-F12, supplemented with 10% FCS and 1% each of glutamine, penicillin, streptomycin and amphotericin B, and then further subcultured. During subculture, keratocyte-derived fibroblasts were detached using a 0.1% trypsin/EDTA solution.

A human corneal endothelial cell line (HTCE) was kindly provided by Pr. K. Engelmann and Dr. J. Bednarz. This cell line was obtained by transfection with the coding gene for the large T proTCECEN and reproduced the morphological and functional characteristics of normal endothelium. It was cultured under standard conditions (5% CO₂, 95% humidified air, 37 °C) in 50% M199 and 50% Ham's F12 supplemented with 10% FCS, 2% glutamine, 50 mg/ml of streptomycin, 50 IU/ml of penicillin, 20 μg/ml ascorbic acid, and 20 μg/ml of insulin.

A human epithelial conjunctival cell line (Wong-Kilbourne derivative of Chang conjunctiva, clone 1-5c-4, ATCC CCL-20.2) was cultured under standard conditions (5% CO₂-95% humidified air, 37 °C) in Eagle's Minimal Essential Medium supplemented with 10% FCS, 2 mmol/l L-glutamine, 50 mg/ml streptomycin, and 50 IU/ml penicillin.

Each cornea led to independent primary cell cultures. Purity and identity of the established cell lines and primary cultures were confirmed using established markers in immunological or reverse transcriptase-polymerase chain reaction (RT-PCR) assays as previously described [22]. Molecular and immunological assays were always performed at the fifth subculture of all the cultures. Homogeneous state of confluence (80%) was also respected.

Reverse transcriptase-polymerase chain reaction experiments

Total mRNA was extracted from human total cornea, conjunctiva and cell cultures using trizol (Invitrogen, Cergy-Pontoise, France). RNA quality was studied via the RNA-to-protein ratio (260 nm/280 nm) and by gel electrophoresis (2% agarose) to observe the presence of intact 28S and 18S RNA bands. Specific oligonucleotide primers were originally generated using the web program Primer3, based on the published full-length human mRNA sequences of each specific gene and designed to avoid genomic DNA amplification (Table 1). All the primers were first checked for their specific ability to amplify defined mRNA regions using human tissue already reported to express these genes (positive controls).

cDNA was generated using the Superscript First-Strand Synthesis System for RT-PCR (Gibco-BRL, Cergy-Pontoise, France). PCR amplification was performed in an Eppendorf Mastercycler (Eppendorf, Le Pecq, France), using 50 ng of total mRNA per reaction and according to the following program: initial denaturation at 95 °C for 5 min followed by denaturation at 95 °C for 45 s, annealing at 59 °C for 45 s, and extension at 72 °C for 60 s (36 cycles), and a final extension of 72 °C for 7 min. The PCR products were separated on 2% agarose gel and sequenced on both strands to confirm the specificity of the reaction using the same primers on a DNA Dye Terminator Cycle Sequencing Kit (Applied Biosytems, Courtaboeuf, France) and the Applied Biosystems model 377 DNA Sequencer. Amplification of the housekeeping gene, acidic ribosomal phosphoprotein P0 36B4 (219 base pairs), was used as positive control. A negative control for amplimer contamination was set up using a complete PCR reaction mix without cDNA.

Immunohistological and immunocytological experiments

Cryosections of total cornea, conjunctiva, and cells grown in Laboratory-Tek chambers (MC2, Clermont-Ferrand, France) were fixed in 4% paraformaldehyde in PBS (pH 7.4) at 25 °C for 10 min, rinsed three times with PBS, incubated at 25 °C for 10 min in H₂O₂ (quenching of endogenous peroxydases), and incubated in PBS with 3% BSA (Sigma Aldrich, St. Quentin Fallavier, France) at 25 °C for 30 min. Cells and tissues were incubated overnight at 4 °C in the presence of RAR α, β, and γ as well as RXR α, β, and γ (Santa-Cruz 551, 552, 550, 553, 831, 555, respectively) and certified rabbit polyclonal primary antibodies (1/200 in PBS; Tebu, Le Perray-en-Yvelines, France). This step was followed by three PBS washes, a 1 h incubation in the presence of a secondary FITC-labeled anti-rabbit antibody (Interchim, Montluçon, France) at room temperature, and three more washes with PBS. The samples were then examined after DAPI nuclear staining (five min, dilution in PBS 1/500) and after mounting in a Vectashield aqueous mounting fluid (Vector, Burlingame, CA) under a Zeiss Axiophot microscope. For negative controls, sections were incubated with normal rabbit IgG in place of the anti-RARs and anti-RXRs.

Transfection of cultured cells

Plasmid DR5-tk-CAT contains one copy of the retinoic acid-responsive element DR5 (direct repeat 5) bound to a herpes simplex thymidine kinase promoter upstream of a chimeric chloramphenicol acetyl transferase (CAT) reporter gene [23]. The human corneal epithelial cells were trypsinized 16 h before transfection. A total of 3x10⁵ cells in six well plates were transfected using GeneJammer with 1 μg of reporter DR5-tk-CAT plasmid and 1 μg of cytomegalovirus (CMV)-β-galactosidase vector serving as internal control to normalize variations in transfection efficiency. The CMV-β-galactosidase plasmid contains CMV promoter and enhancer sequences that drive a β-galactosidase (βGAL) gene. Parallel transfections of the corresponding vectors CAT and β-GAL at equivalent concentrations were performed in all experiments. After overnight incubation with DNA, the cells were washed and incubated with fresh medium for a further 12 h. They were treated for another 24 h with retinoids (retinol or all-trans retinoic acid [ATRA]) in 1000X stock solutions in DMSO. For this step, the maximal DMSO concentration to which the cells were exposed was <0.1%. In parallel experiments, ethanol, bisdiamine (Acros organics, Noisy-le-Grand, France), 4-methylpyrazole and DMSO were added at concentrations of 100 mmol/l, 100 μmol/l, 2 mmol/l, and 500 mmol/l, respectively. Ethanol, bisdiamine, and 4-methylpyrazole were chosen because they are well known to inhibit retinol conversion into retinoic acid. Cell viability assays were performed for each treatment (ethanol, bisdiamine, 4-methylpyrazole, and DMSO) using 3'-[1-[(phenylamino)-carbonyl]-3,4-tetrazolium]bis(4-methoxy-6-nitro) benzensulfonic acid hydrate (XTT) assays (Roche Diagnostics, Meylan, France). For this purpose, cells were seeded on 96 well plates at 5x10⁴ cells/well and treated for 24 h with ethanol, bisdiamine, 4-methylpyrazole, or DMSO at concentrations ranging from 10-200 mmol/l, 10-200 μmol/l, 1-10 mmol/l, and 10-900 mmol/l, respectively. The averages for cells grown in the regular medium were considered as 100% viability, and no toxicity >10% was observed for ethanol between 10 and 100 mmol/l, bisdiamine between 10 and 100 μmol/l, 4-methylpyrazole treatment between 1 and 5 mmol/l, or DMSO treatment between 10 and 500 mmol/l.

Chloramphenicol acetyl transferase and β-galactosidase reporter gene assay

After incubation, human epithelial corneal cells were washed twice with PBS and treated with 700 μl of cell lysis buffer for 1 h at 4 °C. The lysed cells were scraped off and centrifuged at 950 g for five min at 4 °C. CAT was measured by immunoenzymatic assay (Roche Diagnostics, Meylan, France) on 200 μl of supernatant. In all the studies, CAT assays were normalized to β-galactosidase activity, which was determined according to the manufacturer's instructions. For this purpose, 20 μl of the supernatants were incubated with assay reagent based on an original protocol.

Statistical methods

Results, expressed as mean±SD were an average of 10 different experiments per condition. Comparison of means was performed by ANOVA and Fisher's t-test using Statview II 1.03 software (Abacus Concepts Inc., Berkeley, CA). For all the studies conducted, the level of significant difference was set at p<0.05.

Results

Identification of the molecular and metabolic factors of the retinoid pathway in adult human ocular surface

To evaluate the expression patterns of nuclear receptors, binding proteins, and enzymes of the retinoid signaling pathways in human adult ocular surface, RT-PCR assays were first performed on total mRNA isolated from five corneas and ten conjunctivas. Reproducible results were obtained in both tissues (Table 2). RAR α, β, γ, and RXR α are expressed in both the cornea and conjunctiva. RXR β and γ were never detected in the cornea and conjunctiva. ADH3, ADH4, RALDH1, RALDH3, DHRS4, DHRS9, CRBP1, CRABP1, and CRABP2 were expressed simultaneously in the cornea and in the conjunctiva. ADH1, RALDH2, RALDH4, CYP26A, CYP26B, CYP26C, CRBP2, and epimerase were never detected in human ocular total cornea and conjunctiva. RODH4 was only expressed in conjunctiva and not detected in the cornea (Table 2).

These initial data led us to better determine the expression patterns of the nuclear receptors, binding proteins, and enzymes of the retinoid signaling pathways in human corneal and conjunctival compartments at the cellular level. The major cells constituting the cornea and conjunctiva were analyzed using primary cultures or established cell lines such as the primary cell cultures of human corneal epithelium and corneal keratocytes, human corneal endothelial (HTCE) and epithelial (HCE) cell lines, and human epithelial conjunctival cell line (Chang). RAR α, RAR γ, RXR α, RALDH1, RALDH3, DHRS4, CRBP1, CRABP1, and CRABP2 were commonly expressed in corneal epithelial and endothelial cells as well as in keratocytes (Table 3). RAR β was never detected in corneal keratocytes. ADH3, ADH4, and DHRS9 were only detected in epithelial cells. Reproducible results were obtained for each analyzed type of cells. Both epithelial primary cell cultures and established cell lines (HCE) expressed the same nuclear receptors, enzymes, and binding proteins, suggesting that the HCE cell line could be used as a model for studying the biological roles of retinoids in the corneal epithelium. The molecular and metabolic actors previously detected in total conjunctiva (RAR α, RAR β, RAR γ, RXR α, ADH3, ADH4, RODH4, DHRS4, DHRS9, RALDH1, RALDH3, CRBP1, CRABP1, and CRABP2) were all found in human conjunctival epithelial cells.

Using frozen sections, we checked the protein expression and localization of RAR α, β, γ, and RXR α. The transcripts of these had been previously described. We found the four nuclear receptors expressed in the cornea and the conjunctiva (Figure 2A,C, and data not shown). The presence of RAR α, β, γ, and RXR α immunolabeling was also confirmed by analysis of the primary cultures of corneal epithelial and stromal cells and of corneal endothelial and conjunctival established cell lines (Figure 2E,G,H,J,K,L,N,O, and data not shown). Taken together, all the immunocytological and immunohistochemical assays confirmed the presence of the four nuclear receptor proteins, RAR α, β, γ, and RXR α, previously identified by RT-PCR.

Study of retinoic acid generation from retinol by corneal epithelial cells

The production of active retinoids by corneal epithelial cells after incubation with retinol was studied using transient transfections of a retinoid-sensitive reporter gene based on CAT expression and using the HCE cell line (Figure 3). First, we checked that the naive corneal epithelial cells (or corneal epithelial cells transfected with an empty vector) did not produce CAT. CAT production by cells transfected with the DR5-tk-CAT plasmid and not treated was considered as the baseline for detection of and comparison with recorded induction levels following retinoid exposure. All-trans RA treatments (1 μmol/l) for 24 h increased CAT response (1.8±0.2 fold induction), demonstrating the cellular functionality of retinoic acid signaling at the cellular level. CAT production was also significantly enhanced (1.4±0.1) after retinol (1 μmol/l) treatment but not as much as after stimulation with 1 μmol/l of RA. In addition, we found that epithelial corneal cells only generated the all-trans isomer of the retinoic acid from retinol (unpublished data obtained from HPLC experiments). To confirm that active retinoids were generated and catabolized by enzymatic conversion from retinol, we tested potential impairment of these reactions by inhibiting ADH and RALDH, respectively, using ethanol, bisdiamine, and 4-methylpyrazole (at nontoxic concentrations; see Methods). The absolute values of the internal standard (not dependent upon the enzymatic conversion caused by ADH and RALDH) did not change in response to ethanol, bisdiamine, or 4-methylpyrazole treatments (data not shown). This result shows the specific effect of ethanol, bisdiamine, and 4-methylpyrazole on the inhibition of retinoic acid generation and degradation. Ethanol (100 mmol/l), bisdiamine (100 μmol/l), and 4-methylpyrazole (2 mmol/l) all caused a statistically significant decrease in retinoic acid production by primary epithelial corneal cells as illustrated by the insignificant amount of CAT production between control cells and cells treated with retinol and one of the three inhibitors.

Discussion

It has been extensively demonstrated that the cornea and the conjunctiva have an absolute requirement of vitamin A (retinol) thus establishing vitamin A deficiency as an important public health problem in many developing countries and triggering a spate of studies on the clinical uses of retinoids in ocular surface pathologies [5]. Curiously, partial characterization of the retinoid molecular and metabolic actors (RARs and RXRs) have been realized in rabbit, mouse, and human conjunctiva and cornea [2,15,16]. However, there has not yet been reports of systematic tissue and cellular expression patterns of these receptors nor have the metabolic activities on retinoids at the human ocular surface been demonstrated.

Therefore, for the first time, we established a complete expression pattern of RARs and RXRs in human cornea and conjunctiva and in the main cell types. Our data could be put to use to choose specific agonists and antagonists for RARs and RXR α types which could fine-tune a more selective pharmacological approach for treating ocular surface pathologies [24,25]. All three RARs and RXR α were colocalized in conjunctival and corneal tissues and cells, creating an opportunity for the active heterodimer RARs/RXR α to form in the human ocular surface. The establishment of this active heterodimer provides the rationale explaining how retinoic acid regulates previously studied target genes in the ocular surface environment: collagenase [26], α1-proteinase inhibitor [11], interleukin-1α [27], cytochrome P450 (CYP)4B1 [28], and mucins [13,15]. This functional heterodimer also underlies more global physiological effects of the retinoids already described in the cornea such as reduced apoptosis and oxidative stress [29,30] as well as in the conjunctiva (trans-differentiation of epithelial cells [31]). RXR β was present in rabbit and mice eyes but absent from the human ocular surface. This differential expression between the species had to be linked with the absence of phenotype in RXR β null mutant mice, suggesting a weak implication of this nuclear receptor in mammalian ocular physiology. By contrast, the systematic expression of RAR α, RAR γ, and RXR α in all three species (mouse, human, and rabbit) strongly suggests their roles in eye physiology and development, as previously illustrated by the ocular phenotype of the RAR α-, RAR γ-, and RXR α-null mutant mice [32]. Our results concerning RAR β could be considered contradictory to those obtained on human ocular surface by Gipson's group [15]. However, this receptor was found to be expressed in the rabbit cornea and conjunctiva [16]. Furthermore, while certain conflictual results have been found between both conjunctival cell lines, it should be underlined that the concerned studies employed two essentially different cell lines: the telomerase-immortalized human conjunctival epithelial cell line (HCjE) for Hori et al. [13] and the Wong-Kilbourne derivative of Chang conjunctiva, (clone 1-5c-4/ATCC CCL-20.2) for our work, thus establishing the relative heterogeneity of conjunctival cell lines in terms of global mRNA expression patterns.

We have shown for the first time that the corneal epithelial cells are able to produce retinoids at functional levels from retinol, which has already been reported to be present in tears fluid [21]. We established the presence of the cytoplasmic binding proteins (CRBP and CRABP) implicated in this retinoid metabolic pathway. Our results are in agreement with those previously obtained results using radiolabeled retinoid assays, which formed the basis of cytosolic retinoid binding capacities [19,33]. Three other mammalian ocular surface structures have also been shown to produce retinoic acid from retinol: the rabbit [34], the rat [35], and the mouse [36]. For each determination, the induction obtained from retinol was always less than from exposure to all-trans retinoic acid. This could illustrate and explain (1) why Gipson's group did not find the same induction of mucin genes when they used calf serum (containing almost retinol) or retinoic acid diluted in DMSO [15], and (2) why retinoic acid is more efficient than retinol in treating xerophthalmia [37]. Our study also clearly demonstrates that the generation of retinoic acid from retinol can be impaired by a toxin like ethanol. This point will need to be emphasized if retinol is to be used in eye droplets to treat ocular surface pathologies such as xerophthalmia or dry eye. If the galenic form contains ethanol to solubilize the retinol, this could impair its transformation into retinoic acids. The same reflection could also apply if these ocular pathologies were treated by systemic administration of retinol [38,39] as the treatment could be made inefficient by the absence of retinol activation against a background where ethanol is present, such as chronic alcohol addiction. In this scenario, local retinoic treatment (correctly supported [40]) should be preferred. Nevertheless, this topical administration should be strongly followed up. Indeed, we have established an absence of members of the Cyp26 family, which represent one of the main degradation pathways of the retinoic acids thus excluding a physiological capacity to metabolize part of the excess retinoic acid delivery. The absence of such members of the Cyp26 family could also be correlated to potential defects induced by excessive systemic retinoic acid administration during development [41] as well as in adult life [42]. The expression patterns of RA metabolism-associated binding proteins and enzymes found in the corneal endothelial cells suggests that this cellular compartment is also able to produce retinoic acid from retinol. The in situ controlled generation of retinoic acid could modulate the mitogenic effect of epidermal growth factor (EGF) [12] underlining the importance of retinoids produced endogenously in the maintenance of endothelial integrity. These endogenous retinoids could also protect against the free-radical-induced oxidation injury, lipid peroxidation, and consequent endothelial apoptosis [29].

In conclusion, we have identified the metabolic and molecular actors of the retinoid signaling pathway in the adult human ocular surface. Alterations of this metabolic and molecular pathway may be determinant factors in ocular surface physiology and may offer a molecular basis for the explanation and treatment of eye pathologies.

Acknowledgements

F.C., G.M., and V.S. were supported by INSERM grants (Contrat Interface for F.C. and V.S. and Poste Accueil for G.M.). The ARDEMO team received support from the French Research Department (JE 2447). L.B. was supported by a THEA® grant. We thank ATT® for grammatical and vocabulary reading of this manuscript.

References

1. Wilson JG, Roth CB, Warkany J. An analysis of the syndrome of malformations induced by maternal vitamin A deficiency. Effects of restoration of vitamin A at various times during gestation. Am J Anat 1953; 92:189-217.

2. Mori M, Ghyselinck NB, Chambon P, Mark M. Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes. Invest Ophthalmol Vis Sci 2001; 42:1312-8.

3. Hatchell DL, Sommer A. Detection of ocular surface abnormalities in experimental vitamin A deficiency. Arch Ophthalmol 1984; 102:1389-93.

4. Macsai MS, Agarwal S, Gamponia E. Bilateral corneal ulcers in primary vitamin A deficiency. Cornea 1998; 17:227-9.

5. Sommer A, Davidson FR, Annecy Accords. Assessment and control of vitamin A deficiency: the Annecy Accords. J Nutr 2002; 132:2845S-2850S.

6. Ubels JL, Edelhauser HF, Foley KM, Liao JC, Gressel P. The efficacy of retinoic acid ointment for treatment of xerophthalmia and corneal epithelial wounds. Curr Eye Res 1985; 4:1049-57.

7. Herbort CP, Zografos L, Zwingli M, Schoeneich M. Topical retinoic acid in dysplastic and metaplastic keratinization of corneoconjunctival epithelium. Graefes Arch Clin Exp Ophthalmol 1988; 226:22-6.

8. Yoon KC, Im SK, Park YG, Jung YD, Yang SY, Choi J. Application of umbilical cord serum eyedrops for the treatment of dry eye syndrome. Cornea 2006; 25:268-72.

9. Bastien J, Rochette-Egly C. Nuclear retinoid receptors and the transcription of retinoid-target genes. Gene 2004; 328:1-16.

10. Lefebvre P, Martin PJ, Flajollet S, Dedieu S, Billaut X, Lefebvre B. Transcriptional activities of retinoic acid receptors. Vitam Horm 2005; 70:199-264.

11. Boskovic G, Twining SS. Retinol and retinaldehyde specifically increase alpha1-proteinase inhibitor in the human cornea. Biochem J 1997; 322:751-6.

12. Ubels JL, Iorfino A, O'Brien WJ. Retinoic acid decreases the number of EGF receptors in corneal epithelium and Chang conjunctival cells. Exp Eye Res 1991; 52:763-5.

13. Hori Y, Spurr-Michaud SJ, Russo CL, Argueso P, Gipson IK. Effect of retinoic acid on gene expression in human conjunctival epithelium: secretory phospholipase A2 mediates retinoic acid induction of MUC16. Invest Ophthalmol Vis Sci 2005; 46:4050-61.

14. Zhou J, Kochhar DM. Regulation of AP-2 and apoptosis in developing eye in a vitamin A-deficiency model. Birth Defects Res A Clin Mol Teratol 2003; 67:41-53.

15. Hori Y, Spurr-Michaud S, Russo CL, Argueso P, Gipson IK. Differential regulation of membrane-associated mucins in the human ocular surface epithelium. Invest Ophthalmol Vis Sci 2004; 45:114-22.

16. Bossenbroek NM, Sulahian TH, Ubels JL. Expression of nuclear retinoic acid receptor and retinoid X receptor mRNA in the cornea and conjunctiva. Curr Eye Res 1998; 17:462-9.

17. Liden M, Eriksson U. Understanding retinol metabolism: structure and function of retinol dehydrogenases. J Biol Chem 2006; 281:13001-4.

18. Taimi M, Helvig C, Wisniewski J, Ramshaw H, White J, Amad M, Korczak B, Petkovich M. A novel human cytochrome P450, CYP26C1, involved in metabolism of 9-cis and all-trans isomers of retinoic acid. J Biol Chem 2004; 279:77-85.

19. Wiggert B, Bergsma DR, Lewis M, Abe T, Chader GJ. Vitamin A receptors. II. Characteristics of retinol binding in chick retina and pigment epithelium. Biochim Biophys Acta 1977; 498:366-74.

20. Wei RR, Wamer WG, Lambert LA, Kornhauser A. β-Carotene uptake and effects on intracellular levels of retinol in vitro. Nutr Cancer 1998; 30:53-8.

21. Ubels JL, MacRae SM. Vitamin A is present as retinol in the tears of humans and rabbits. Curr Eye Res 1984; 3:815-22.

22. Chiambaretta F, De Graeve F, Turet G, Marceau G, Gain P, Dastugue B, Rigal D, Sapin V. Cell and tissue specific expression of human Kruppel-like transcription factors in human ocular surface. Mol Vis 2004; 10:901-9 <http://www.molvis.org/molvis/v10/a108/>.

23. Mader S, Chen JY, Chen Z, White J, Chambon P, Gronemeyer H. The patterns of binding of RAR, RXR and TR homo- and heterodimers to direct repeats are dictated by the binding specificites of the DNA binding domains. EMBO J 1993; 12:5029-41.

24. Germain P, Chambon P, Eichele G, Evans RM, Lazar MA, Leid M, De Lera AR, Lotan R, Mangelsdorf DJ, Gronemeyer H. International Union of Pharmacology. LX. Retinoic acid receptors. Pharmacol Rev 2006; 58:712-25.

25. Germain P, Chambon P, Eichele G, Evans RM, Lazar MA, Leid M, De Lera AR, Lotan R, Mangelsdorf DJ, Gronemeyer H. International Union of Pharmacology. LXIII. Retinoid X receptors. Pharmacol Rev 2006; 58:760-72.

26. Pirie A, Werb Z, Burleigh MC. Collagenase and other proteinases in the cornea of the retinol-deficient rat. Br J Nutr 1975; 34:297-309.

27. West-Mays JA, Cook JR, Sadow PM, Mullady DK, Bargagna-Mohan P, Strissel KJ, Fini ME. Differential inhibition of collagenase and interleukin-1αgene expression in cultured corneal fibroblasts by TGF-β, dexamethasone, and retinoic acid. Invest Ophthalmol Vis Sci 1999; 40:887-96.

28. Ashkar S, Mesentsev A, Zhang WX, Mastyugin V, Dunn MW, Laniado-Schwartzman M. Retinoic acid induces corneal epithelial CYP4B1 gene expression and stimulates the synthesis of inflammatory 12-hydroxyeicosanoids. J Ocul Pharmacol Ther 2004; 20:65-74.

29. Serbecic N, Beutelspacher SC. Anti-oxidative vitamins prevent lipid-peroxidation and apoptosis in corneal endothelial cells. Cell Tissue Res 2005; 320:465-75.

30. Serbecic N, Beutelspacher SC. Vitamins inhibit oxidant-induced apoptosis of corneal endothelial cells. Jpn J Ophthalmol 2005; 49:355-62.

31. Tseng SC, Farazdaghi M. Reversal of conjunctival transdifferentiation by topical retinoic acid. Cornea 1988; 7:273-9.

32. Kastner P, Grondona JM, Mark M, Gansmuller A, LeMeur M, Decimo D, Vonesch JL, Dolle P, Chambon P. Genetic analysis of RXR αdevelopmental function: convergence of RXR and RAR signaling pathways in heart and eye morphogenesis. Cell 1994; 78:987-1003.

33. Wiggert B, Bergsma DR, Helmsen R, Chader GJ. Vitamin A receptors. Retinoic acid binding in ocular tissues. Biochem J 1978; 169:87-94.

34. Hough RB, Piatigorsky J. Preferential transcription of rabbit Aldh1a1 in the cornea: implication of hypoxia-related pathways. Mol Cell Biol 2004; 24:1324-40.

35. Julia P, Farres J, Pares X. Ocular alcohol dehydrogenase in the rat: regional distribution and kinetics of the ADH-1 isoenzyme with retinol and retinal. Exp Eye Res 1986; 42:305-14.

36. Lee MO, Manthey CL, Sladek NE. Identification of mouse liver aldehyde dehydrogenases that catalyze the oxidation of retinaldehyde to retinoic acid. Biochem Pharmacol 1991; 42:1279-85.

37. Hatchell DL, Faculjak M, Kubicek D. Treatment of xerophthalmia with retinol, tretinoin, and etretinate. Arch Ophthalmol 1984; 102:926-7.

38. Sethi V, Kapil U. Reduction in prevalence of Bitot's spots after administration of mega-dose of vitamin A. Indian Pediatr 2003; 40:585-7.

39. Singh V, West KP Jr. Vitamin A deficiency and xerophthalmia among school-aged children in Southeastern Asia. Eur J Clin Nutr 2004; 58:1342-9.

40. Selek H, Unlu N, Orhan M, Irkec M. Evaluation of retinoic acid ophthalmic emulsion in dry eye. Eur J Ophthalmol 2000; 10:121-7.

41. Evans K, Hickey-Dwyer MU. Cleft anterior segment with maternal hypervitaminosis A. Br J Ophthalmol 1991; 75:691-2.

42. Lebowitz MA, Berson DS. Ocular effects of oral retinoids. J Am Acad Dermatol 1988; 19:209-11.