Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The upper respiratory tract is lined by a pseudostratified epithelium, composed of basal cells, secretory cells, and ciliated cells. Retinoids, vitamin A and its derivatives, are required for the maintenance of the epithelial mucociliary (normal) phenotype. Retinoids exert their effects via specific nuclear receptors, retinoic acid receptors (RARs) RAR, RAR, and RAR and retinoid X receptors (RXRs) RXR, RXR, and RXR, which are members of the steroid/thyroid receptor superfamily. RAR/RXR heterodimers and RXR/RXR homodimers act as ligand-dependent transcription factors, interacting with retinoic acid response elements (RAREs) located in promoter regions of target genes (1, 2). RAR, -, and - and RXR and - receptor proteins and mRNA have been detected in vivo by immunohistochemistry in normal human lung tissue (3). mRNA of all RARs and RXRs has also been identified in normal human oral tissue and in 89% of normal bronchial specimens (4). All receptors have also been detected in vitro by reverse transcriptase/polymerase chain reaction (RT-PCR) analyses (5).

Physical or chemical injuries alter the respiratory epithelium, which becomes stratified as a result of a multistep process, squamous metaplasia (6). This state, displaying similarities with epidermal differentiation, is often considered to be a precancerous lesion (9, 10). In vivo, vitamin A deficiency induces replacement of the normal pseudostratified mucociliary epithelium by a metaplastic stratified squamous epithelium (11, 12), a process that can be reversed by a dietary vitamin A supplement (13). The physiologic effect of vitamin A can be reproduced in in vitro studies using organs or respiratory epithelial cell cultures. In the absence of vitamin A, cells undergo squamous differentiation (14). Retinoic acid (RA) inhibits the expression of squamous-related genes, such as keratin 13 in rabbit tracheobronchial cells (17), cholesterol sulfate in human epithelial cells (18), and transglutaminase I (Tg I) and cornifin in human epidermal keratinocytes (19). RA also induces a mucosecretory phenotype by activating both transcription of specific mucin genes, particularly MUC2, MUC5B and MUC5AC, and mucin secretion (20). In parallel, retinoids are required for the presence of ciliated cells, as demonstrated by cell culture morphological studies. However, the ciliated cell differentiation process has not been quantified. We have previously shown that such a quantification of ciliated cells can be performed using GT335, a monoclonal antibody raised against glutamylated tubulin (23), which specifically recognizes cilia axonemes (24).

In this study, we used GT335 to detect ciliated cells in the air-liquid interface (ALI) culture model and to test the role of retinoids in the ciliated cell differentiation process. We tested the effect of selective agonists for each RAR subtype during in vitro differentiation of human nasal epithelial (HNE) cells. Using glutamylated tubulin (GT335-positive cells) and Tg I as markers of ciliated cells and squamous cells, respectively, we show that agonists selective for the three RAR subtypes induce ciliated cell differentiation and, in parallel, inhibit squamous differentiation. These effects are specifically dependent on RARs, because antagonists specific for the three RARs inhibited the activity of the corresponding agonists. Moreover, addition of RARs-specific agonists restored a ciliated phenotype in retinoid-deficient cultures. These results demonstrate that all three RARs are involved in the response of HNE cells to RA.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture and Treatment Procedure

Human nasal polyps or turbinates were obtained from patients undergoing polypectomy or turbinectomy (Dr. A. Coste, Prof. Peneygre's department, Hôpital Intercommunal de Créteil, France). They were washed in Dulbecco's modified Eagle medium (DMEM)/ F12 (GIBCO BRL, Grand Island, NY) and incubated with 2 mg/ml of pronase (Protease XIV; Sigma, Saint Quentin Fallavier, France) in DMEM/F12 supplemented with 50 U-50 mg/ml of penicillin-streptomycin at 4°C for 16-20 h under rotary agitation (80 rpm). Ten percent fetal calf serum was then added to neutralize the enzyme. After washing, aggregates were discarded and dissociated cells were preplated for 2 h at 37°C on plastic dishes (Falcon Merck-Eurolab, Strasbourg, France) to eliminate most contaminating fibroblasts. The cell suspension was then filtered on a 30-µm diameter filter, and epithelial cells were counted. After another centrifugation at 400 × g for 5 min, cells were resuspended in a 1/1 BEGM/DMEM/F12 (Clonetics [Biowhittaker, Emerainville, France], Gibco) mixture supplemented with insulin (Clonetics, 5 µg/ml), hydrocortisone (Clonetics, 0.5 µg/ml), epinephrine (Clonetics, 0.5 µg/ml), triiodothyronine (Clonetics, 6.5 ng/ml), transferrin (Clonetics, 10 ng/ml), human epidermal growth factor (Clonetics, 0.5 ng/ml), bovine pituitary extract (Clonetics, 0.13 mg/ml), gentamicin:amphotericin (Clonetics, 50 µg/ml:50 ng/ml) and bovine serum albumin (BSA) (Sigma, 1.5 µg/ml). Cells were plated in 300 µl medium at a density of 3.10⁴ cells per cm² onto type I collagen (Human placenta collagen)-coated semipermeable membranes (Transwell; Costar, Dominique Deutscher, Brumath, France). Seven hundred microliters of medium were deposited on the basolateral side. Medium was changed every two days. Cells were grown submerged until confluence. At confluence (day 0), the ALI was created and the cultures were treated with retinoids. RA was purchased from Sigma; and the RAR-selective agonists were provided by Galderma (Sophia-Artipolis, France). RA and synthetic retinoids were dissolved in DMSO. The RAR-selective agonists used were CD336, CD2019, and CD666 for RAR, RAR, and RAR respectively. kD values for each receptor are listed in Table 1. kD values have been determined by competition experiments with labeled pan-agonist CD367 as demonstrated by Martin and colleagues (25). CD2503 and CD2665 were used as RAR and RAR/ antagonists, respectively. Control cultures were treated with DMSO only. For the reversion experiments, cells were treated 15 d after creation of the ALI. Cultures were maintained at 37°C in a 5% CO₂ atmosphere.

Antibodies

GT335 is a monoclonal antibody raised against a chemically glutamylated C-terminal peptide of -tubulin (23). Anti-Tg I is a monoclonal antibody B.C1 directed against human keratinocyte transglutaminase (Biomedical Technologies, Inc., Soughton, MA). Anti-MUC5AC is a monoclonal antibody purchased from Neomarkers (Fremont, CA).

Flow Cytometry

Cells were dissociated at the indicated times and were centrifuged at 500 × g for 5 min. They were resuspended in cell dissociation buffer (Sigma) and fixed in MeOH at 20°C. Cells were centrifuged at 2700 × g, and the pellets were washed in phosphate-buffered saline (PBS) containing 0.1% Tween 20. Cells were incubated with 100 µl of MUC5AC (1/100) diluted in 0.1% Tween 20, 1% BSA in PBS for 1 h. The suspension was centrifuged after addition of 1 ml of 0.1% Tween 20 in PBS. Cells were incubated with 100 µl of fluorescein isothiocyanate (FITC)-coupled secondary antibody diluted in 0.1% Tween 20, 1% BSA in PBS for 30 min. Cells were washed with 0.1% Tween 20 in PBS and pellets were incubated with RNase (1 µg/ml) and propidium iodide (20 µg/ml) diluted in PBS. The cell suspension was then analyzed on an Epics-Elite cytometer (Elite EST, Beckman Coulter, Paris Nord-Villepinte, France). An Argon laser was used at an excitation wavelength of 488 nm. FITC fluorescence was detected at 525 nm and phosphatidylinositol fluorescence was detected at 620 nm. Forward angle scatter (FSC) represents cell size and right angle scatter (SSC) corresponds to cell granulometry, i.e., the cell contents.

Protein Extraction

Cells were dissociated with 0.05% trypsin-0.02% ethylenediaminetetraacetic acid (Gibco) in DMEM/F12 for 15 min at 37°C. The action of trypsin was inhibited by adding 10% fetal calf serum. Cells were collected and centrifuged at 500 × g for 5 min at room temperature. Pellets were resuspended in cell dissociation buffer. After centrifugation at 500 × g, pellets were resuspended in Laemmli buffer (1×) containing 1 mM phenylmethylsulfonyl fluoride. An aliquot was taken for the assay. Protein samples were quantified using the bicinchroninic acid kit (Pierce, Perbio Science France, Bezons, France). Following addition of 5% beta mercaptoethanol, the samples were boiled and then frozen at 20°C.

Electrophoresis and Immunoblots

For one-dimensional electrophoresis, protein extracts resuspended in Laemmli buffer 1 × were boiled for 5 min and loaded onto a 10% polyacrylamide gel according to Laemmli's technique (26). Proteins were transferred from the gel onto nitrocellulose membranes according to the technique of Towbin and colleagues (27). After staining with Ponceau red, the blots were saturated with Tris-buffered saline (TBS) containing 5% milk and 0.1% Tween 20 for 1 h at room temperature. They were incubated with primary antibody diluted in TBS 0.1% Tween 20 and 1% milk for 1 h at room temperature. The blots were washed in TBS 0.1% Tween 20 and incubated with the peroxidase-labeled secondary antibody for 30 min at room temperature. After intensive washing, detection was performed according to the chemoluminescence technique (ECL; Amersham, Pharmacia Biotech, Europe GmbH, Orsay Cedex, France).

Scanning Electron Microscopy

Cultures were rinsed in PBS and in 0.1 M cacodylate buffer, fixed for 1 h in 2% glutaraldehyde in 0.05 M cacodylate buffer (pH 7.4), rinsed for 1 h in 0.1 M cacodylate buffer, and then postfixed in 1% OsO₄ in 0.05 M cacodylate buffer. The samples were then dehydrated in a graded ethanol series. The samples were embedded in Epon (Taab, Delville Technology, St. Germain-en-Lay, France) or stored in acetone and critical-point dried using CO₂ in a Balzers apparatus (Boizian Distribution, Selles sur cher, France). After being coated with a gold layer, samples were examined by a JEOL 6100 scanning electron microscope (Jeol).

RT-PCR

Total RNA was isolated from pools of triplicate cultures by collecting cells in RNAble (Eurobio, Les Ulis, France). Following addition of 10% chloroform, samples were incubated for 5 min at 4°C and centrifuged for 10 min at 18,000 × g at 4°C. RNAs contained in the aqueous phase were precipitated by adding 2 vols of ethanol for at least 24 h at 20°C. RNAs were then pelleted by centrifugation for 30 min at 18,000 × g at 4°C, rinsed in 70% ethanol, centrifuged again, and resuspended in diethylpyrocarbonate-treated water (Croissy sur Seine, France). After assay, 10 µg of RNA were reverse-transcribed using OligodT Primer and Superscript II RNase H Reverse Transcriptase (both from Gibco). PCR reactions were performed on 1 ng of cDNA using the Expand Long Template PCR system (Roche Molecular Biochemicals, Meylan, France) in a PTC-100 thermocycler (MJ Research, Inc., Watertown, South Dakota). The amplification conditions were as follows: between 25 and 40 cycles of denaturation (95°C, 1 min), annealing (58°C, 1 min), and extension (68°C, 1 min), depending on gene products, in the presence of 3 mM MgCl₂ and 0.3 µM primers. The sequences used for each primer were as follows: GAPDH, 5' primer: ACC ACA GTC CAT GCC ATC AC, 3' primer: TCC ACC ACC CTG TTG CTG TA; Tg I, 5' primer: GCG GCA GGA GTA TGT TCT TA, 3' primer: AAG GGA TGT GTC TGT GTC GTG; RAR, 5' primer: GTC TTT GCC TTC GCC AAC CA, 3' primer: GCC CTC TGA GTT CTC CAA CA; RAR, 5' primer: GGA ACG CAT TCG GAA GGC TT, 3' primer: GGA AGA CGG ACT CGC AGT GT; RAR, 5' primer: TGC GAA ATG ACC GGA ACA AG, 3' primer: CCC AGC AAA GGC AAA GAC AA. PCR products were analyzed by electrophoresis on 1.5% agarose gel stained with ethidium bromide. Densitometric analyses were performed using Bio1D software (Vilber Lourmat, Marne-la-Vallée, France). Band intensity was normalized with respect to the GAPDH mRNA content.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of RA on Epithelial Differentiation

Cells dissociated from human nasal polyps or turbinates were grown in RA-supplemented or RA-deprived medium. We analyzed the epithelial cell differentiation process by morphologic and molecular approaches. Ultrastructural examinations of RA-treated cultures at day 21 revealed the presence of numerous secretory and ciliated cells (Figure 1A). MUC5AC gene product, a major mucin in respiratory epithelial cells (28, 29), was detected using a monoclonal anti-MUC5AC antibody, and MUC5AC-positive cells were quantified by flow cytometry during differentiation (Figure 1B). At confluence, only a few cells expressed MUC5AC independently of the presence of RA during epithelial cell proliferation. In RA-treated cultures, the percentage of MUC5AC-positive cells increased with time (Figure 1B). The absence of RA led to morphological and biochemical modifications of the epithelial cells, which displayed a squamous-like phenotype. Epithelial cells were stratified, devoid of ciliated cells, and displayed large, flat squamous epithelial cells at the surface (Figure 1A). Under these conditions, the percentage of MUC5AC-positive cells remained low during differentiation (Figure 1B). Nevertheless, two cell populations were detected at day 17, one of them expressing low levels (23%) of MUC5AC. The morphological characteristics of RA-deprived and RA-treated cell cultures were analyzed by flow cytometry at day 17 (Figure 1C). In contrast with the relatively homogeneous RA-treated cell population, the range of size (FSC) and granulometry (SSC) was significantly higher in the RA- deprived cell population.

View larger version (37K): [in this window] [in a new window]   Figure 1.   Effect of RA on airway epithelial cell differentiation. HNE cells were grown in RA-sufficient (+RA) or RA-deficient medium (RA), and cultured at ALI from confluence (day 0). (A) Primary cultures of HNE cells 21 d after confluence (scanning electron microscopy). RA-treated (+RA) cultures display a mucociliary phenotype with fully differentiated ciliated and secretory cells. By contrast, RA-deprived cells (RA) are stratified, flat and desquamating. The calibration bar represents 10 µm. (B) Measurement of the percentage of MUC5AC-positive cells by flow cytometry during differentiation. At indicated times (0, 7, 17 d after confluence), RA-treated (+RA) and untreated (RA) cells were dissociated, fixed and stained with an anti-MUC5AC monoclonal antibody. Panels represent cell size (FSC) as a function of FITC (MUC5AC-positive) staining. In RA-treated cultures, the percentage of MUC5AC-positive cells increased with time, whereas it remained low in RA-deprived cultures (< 8%) except for a population of large cells (23%) weakly positive for MUC5AC at Day 17. (C) Different patterns of the two cell populations at Day 17 (FSC, cell size; granularity, SSC).

The ciliogenesis process was followed by Western blot using GT335, a monoclonal antibody raised against glutamylated tubulin, which specifically recognizes basal bodies and ciliary axonemes of ciliated epithelial cells (24). Tg I, a protein specifically induced during squamous differentiation in keratinocytes, was chosen as a marker of squamous differentiation of respiratory epithelial cells.

Ciliogenesis was detected when RA was added to the medium, as revealed by the increased signal intensity obtained with GT335 from Day 17 (Figure 2A). By contrast, without RA, no signal was detected with GT335, but the expression of Tg I protein and mRNA increased progressively during differentiation (Figures 2A and 2B). Accordingly, Tg I protein and mRNA expression was downregulated by RA, as determined by Western blot and semi-quantitative RT-PCR assays (Figures 2A and 2B). For example, ~ 6-fold inhibition of mRNA expression was observed at day 14 (Figure 2B).

View larger version (36K): [in this window] [in a new window]   Figure 2.   Effect of RA on squamous and ciliated differentiation markers expression. (A) Immunoblot analysis of total protein extracts of HNE cells using GT335 and an anti-Tg I antibody during differentiation. RA inhibits Tg I expression and induces ciliogenesis. Ten µg of proteins were loaded on each lane. (B) Total RNA was collected at different times from control and RA-treated cultures and the Tg I mRNA level was determined by semiquantitative RT-PCR. GAPDH gene was used as a control gene, not affected by RA treatment. PCR products were run on Ethidium Bromure-stained agarose gels and detected by UV illumination. The histogram represents densitometric analysis of the downregulating effect of RA on Tg I mRNA expression.

The Three RAR Agonists Enhanced Mucociliary Differentiation

All RAR isotypes are continuously expressed in HNE cells during differentiation in vitro (Figure 3A). In all cases, the three RARs are expressed at Day 14, corresponding to the time of treatment in reversion experiments (see below). Cultures were continuously treated since confluence with CD336, CD2019, and CD666, selective agonists for RAR, RAR, and RAR, respectively, at different concentrations. The expression of differentiation markers was assessed by Western blot analyses (Figure 3B). A gradual effect was observed with the three agonists from 0.1 to 10 nM. Little or no effect was observed at 0.1 nM. At concentrations of 1 nM and higher, the agonists induced a decrease of Tg I expression and an increase of the GT335 signal compared with control conditions, and both of these effects were more pronounced after treatment with 10 nM retinoids.

View larger version (42K): [in this window] [in a new window]   Figure 3.   RAR-selective agonist effects on epithelial differentiation. (A) RAR isotypes expression in HNE cells. The three RARs are expressed during differentiation in the presence and in the absence of RA, 107 M (RT-PCR). PCR cycles (40, 40, and 30) were performed for RAR, RAR, and RAR, respectively. (B and C) HNE cells were seeded on collagen matrix in retinoid-deficient medium. At confluence, ALI was created and retinoids (CD336, CD2019, and CD666, agonists selective for RAR, RAR, and RAR, respectively) were added to the medium. Control cultures were treated with DMSO only. Protein and mRNA samples were prepared during differentiation. (B) Retinoids were used at 0, 0.1, 1, or 10 nM. The three RAR-selective agonists inhibited Tg I protein expression and induced ciliogenesis in a dose- and time-dependent manner. Ten micrograms of proteins were loaded on each lane. (C) Histogram representing densitometric analysis of Tg I mRNA expression. Total RNA was collected at the indicated times from control and retinoid-treated cultures, and Tg I and GAPDH mRNA levels were determined by semiquantitative RT-PCR. All three retinoids clearly downregulated Tg I mRNA expression.

Tg I mRNA expression was further analyzed by RT-PCR. All three retinoids, used at concentrations close to their respective Kd, inhibited Tg I mRNA expression (Figure 3C). mRNA level and protein expression were therefore affected by the three retinoids. The three agonists induced the same effect on differentiation, by both downregulating squamous-specific marker expression and inducing a ciliated phenotype, suggesting that the three RARs share several common targets as transcription factors in HNE cells.

Effects of RAR-Selective Agonists Depend on Their Specific Nuclear Receptors

To determine whether the effects of retinoids are mediated by their specific receptors, RAR agonists and antagonists were tested alone and in combination. CD2503 and CD2665, antagonists for RAR and RAR/, respectively, were used at 10-nM and 100-nM. Tubulin glutamylation, and Tg I expression was compared at day 21 (Figure 4A). The two antagonists, CD2503 and CD2665, alone induced an increased Tg I protein expression compared with control conditions, suggesting that a small amount of retinoids was present in the culture medium. In association with CD336 (1 nM), the corresponding antagonist CD2503 had a partial inhibitory effect on CD336 activity at 10 nM (data not shown). At a higher concentration (100 nM), CD2503 totally abolished the CD336 effect, because no ciliated cells were detected (Figure 4A). Tg I expression also decreased with the combination of both CD2503 and CD336. The effects of CD2019 and CD666 were also totally abolished using the RAR/ antagonist CD2665 at 10 nM and 100 nM, respectively. These results show that the effects of the three RAR agonists on epithelial cell differentiation are mediated via an RAR-dependent pathway.

View larger version (53K): [in this window] [in a new window]   Figure 4.   (A) RAR-dependent effect of retinoids on epithelial differentiation: competition with selective antagonists. Epithelial cells were treated with RAR-selective agonists and antagonists, alone or in combination. Glutamylated tubulin (GT335) and Tg I expression were analyzed 21 d after confluence. The effects of CD336, CD2019, and CD666 are abolished by their specific antagonists. (B) Epithelial cells were grown in retinoid-free medium for 15 d after confluence (squamous differentiation). At Day 15, cells were treated with CD336 (10 nM), CD2019 (50 nM), and CD666 (50 nM). Retinoids induced partial reversion of the squamous differentiated phenotype (downregulation of Tg I expression) toward a ciliated phenotype (increased signal with GT335).

RAR-Selective Agonists Can Reverse the Differentiation Process In Vitro

To test the reversible effect of the three RAR-selective agonists, squamous-like epithelial cells were treated at day 15. Under control conditions, no ciliated cells were detected and Tg I protein expression was maintained (Figure 4B). At Day 15, squamous differentiated cultures were then continuously treated with CD336, CD2019, or CD666. The agonists of the three RARs inhibited Tg I expression and, in parallel, induced partial ciliated cell differentiation. These experiments show that, in vitro, RAR-selective agonists are able to reverse a squamous-like phenotype into a normal phenotype.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

RA exerts various biological effects involved in the control of cell proliferation, differentiation, and apoptosis (30). In respiratory epithelial cells, RA is involved in the maintenance of the mucociliary phenotype. In vivo studies have demonstrated that vitamin A deficiency leads to the replacement of the mucociliary epithelium by a squamous epithelium. In vitro, RA inhibits the expression of several proteins that are markers of the squamous phenotype, such as keratin 13 (17, 31), or relaxin (32). Other proteins related to the squamous phenotype in epidermis are also inhibited by retinoids, particularly cornifin (33), involucrin (34), and Tg I (19). RA has also been shown to induce expression and secretion of specific mucins by human respiratory epithelial cells (5, 21).

In this study, we investigated the regulation of HNE cell differentiation by RAR-selective agonists. Tg I was chosen as a marker of squamous differentiation and GT335, a monoclonal antibody raised against glutamylated tubulin (23), was used as a marker of ciliated cells (24). The RA-dependent pattern of expression and secretion of mucins has been documented (22). Nevertheless, the effect of RA on the ciliated cell differentiation process has only been studied at the morphological level. In ciliated epithelial cells from many species, tubulin polyglutamylation has been shown to be restricted to basal bodies and ciliary axonemes of ciliated cells (24, 35). Thus, GT335 specifically recognizes ciliated cells and allows their quantification within a cell population by flow cytometry and Western blot (24). In the absence of a human respiratory epithelial cell line able to differentiate in vitro, we used primary cultures of HNE cells. An efficient way to obtain mucociliary differentiation in vitro is the use of the culture system initially described by Jorissen and colleagues (36). Nevertheless, the ALI model offers the possibility to favor either mucociliary differentiation, i.e., normal differentiation, or squamous differentiation (16). Nasal epithelial cells cultured in RA-deficient medium undergo metaplastic squamous differentiation, characterized by stratification of the epithelium. By contrast, RA induces morphologic modifications leading to a pseudostratified and mucociliary epithelium: a progressive increase in the proportion of secretory and ciliated cells can be observed and quantified (see Figures 1 and 2). RA therefore stimulates mucin mRNA expression (37) and mucin secretion (20, 21), and increases the percentage of secretory cells from HNE cells (our results). Biochemical and molecular analyses show that RA inhibits the expression of Tg I at the protein and mRNA levels. This has been previously observed in rabbit tracheal epithelial cells (31, 38), human head and neck squamous carcinoma cell lines (39) and human keratinocytes (19, 38, 40). RA has also been shown to downregulate both Tg I mRNA expression and Tg I enzymatic activity in human keratinocytes (38).

Because all three RARs are expressed in HNE cells in vitro (see Figure 3A), we tested the effects of agonists selective for each RAR isotype to determine the implication of these receptors in RA effects on epithelial differentiation. Selective agonists for the three RARs were able to mimic the effects of RA, including inhibition of Tg I protein and mRNA expression and induction of ciliated cell differentiation. The three retinoids exerted a dose-dependent effect, being effective at a very low concentration (1 nM) for both inhibition of Tg I expression and induction of ciliated cell differentiation. Whereas the mucosecretory process appears to be preferentially mediated via RAR and to a lesser extent via RAR (5), ciliogenesis may therefore be mediated via each RAR subtype. In the ciliated cell differentiation process, RARs could exert redundant effects for induction of ciliogenesis, while other specific events may more specifically involve one type of receptor, as already demonstrated for the secretory process. Consequently, the specificity of a particular receptor could explain why all receptor isoforms are expressed in nasal epithelial cells.

RAR-selective agonists induce ciliated cell differentiation and inhibit squamous cell differentiation. We have shown here that they are also able to induce conversion of squamous epithelium into ciliated epithelium (see Figure 4), because treatment of squamous RA-deficient cultures with one of the three RARs induced restoration of a ciliated epithelium, characterized by the absence or decreased levels of Tg I protein expression in parallel with increased levels of glutamylated tubulin. Koo and colleagues (20) showed that RA is able to reverse a squamous phenotype toward a mucosecretory phenotype via a series of sequential events, consisting of a decrease in cornifin  gene expression followed by induction of mucin gene expression. Nevertheless, it is reasonable to suppose that in the context of ciliated cell differentiation, inhibition of Tg I expression is not a prerequisite to induction of ciliogenesis, because the two processes may concern different epithelial cells in a cell population. Despite their selectivity for specific receptors, synthetic retinoids can mediate their effects independently of these receptors, probably via orphan receptors. For example, CD437, an RAR agonist, induces apoptosis in a few cell types via an RAR-independent pathway (41, 42). In the present study, we show that RAR-selective agonists act via their specific receptors in the differentiation process, because their action is inhibited by their corresponding antagonists (see Figure 4).

The molecular mechanisms involved in mucociliary differentiation are unknown. RA effects suggest that this process probably involves transcriptional activation of genes containing RAREs in their promoter region. However, it could also involve indirect regulatory mechanisms, because MUC2 gene expression, which does not possess RARE (43, 44), is upregulated by retinoids in human epithelial respiratory cells. Identification of specific gene targets and characterization of other transcription factors involved in ciliated cell differentiation will provide potential keys to the understanding of the biochemical events involved in this complex process. From this study, we conclude that all three RARs can participate in the response of epithelial cells to RA, both by inducing ciliated cell differentiation and by inhibiting squamous cell differentiation.