Introduction

Top Abstract Introduction Methods Results Discussion References

Topical all-trans RA treatment causes an increase in the number of living cell layers in the epidermis of both human and mouse (Connor et al., 1986; Fisher et al., 1991; Griffiths et al., 1993). Hyperplasia results from an induction of cell proliferation in the basal layer of epidermis within 24 hr, as measured by DNA synthesis (Lützow-Holm et al., 1994; Connor and Lowe, 1983). In the hairless mouse, this effect is induced both by active metabolites of all-trans RA and by synthetic retinoid analogs (Connor et al., 1986; Reynolds et al., 1993). Hyperplasia can also be induced by treatments that compromise the epidermal barrier or interfere with integrity of the stratum corneum (Denda et al., 1996). Several studies suggest that the induction of epidermal hyperplasia by all-trans RA and, by implication, other retinoids may be a nonspecific or irritant effect (Fisher et al., 1991; Lützow-Holm et al., 1995; Marks et al., 1990) analogous to that induced by topical abrasives and detergents. Paradoxically, retinoids also have well-characterized antiproliferative effects in epidermis. Retinoids inhibit epidermal ODC activity and an initial phase of thymidine incorporation after treatment with TPA in mouse (Gendimenico et al., 1989; Verma et al., 1979), and in humans, retinoids are effective both topically and systemically in the treatment of psoriasis, a hyperproliferative disease (Ellis and Voorhees, 1987; Esgleyes-Ribot et al., 1994).

The primary targets for retinoid regulation of gene expression are nuclear receptors from the steroid receptor superfamily. Based on sequence homology and ligand binding specificity, the receptors fall into two groups: the RARs, which bind both all-trans RA and 9-cis RA with high affinity, and the RXRs, which bind 9-cis but not all-trans RA (Mangelsdorf et al., 1994). Synthetic retinoids now also distinguish the two receptor classes (Beard et al., 1995; Lehmann et al., 1992). RAREs in retinoid-inducible genes interact with a heterodimer of RAR and RXR. Both RAR and RXR are in direct contact with DNA (Kurokawa et al., 1994), and the heterodimer appears to be the ultimate nuclear target of all-trans RA induction of gene expression (Mangelsdorf et al., 1994). In addition to heterodimerizing with the RARs, the RXRs are required partners for several other nuclear receptors, including the thyroid hormone, vitamin D₃ and PPAR receptors (Mangelsdorf et al., 1994). Mouse and human skin express a subset of the retinoid nuclear receptors, the RARs  and  and RXR (Elder et al., 1991; Fisher et al., 1994; Kumar et al., 1994). For all-trans RA induction of epidermal hyperplasia, several mechanisms seem to be possible: (i) binding to RARs, (ii) binding to RXRs after isomerization to 9-cis RA and (iii) nonspecific disruption of epidermal integrity.

To determine the role of receptor binding in induction of epidermal hyperplasia, retinoids with defined pharmacological activity (Beard et al., 1995; Lehmann et al., 1992; Nagpal et al., 1995) were evaluated for potency in hyperplasia assays in the hairless mouse. Mutation of the hairless gene on chromosome 14 leads to hair loss after the first cycle of hair growth (Begona et al., 1994), but the hairless mouse has a normal retinoid receptor complement in skin (Beehler et al., 1995). We find that retinoid-induced epidermal hyperplasia is primarily an effect of binding to the RAR class of receptors and is unrelated to retinoid detergent-like character as measured by cell culture cytotoxicity (Oda et al., 1996). In addition, synthetic ligands selective for the RXRs have a greatly reduced potency for induction of hyperplasia and topical irritation compared with RAR-specific retinoids. Finally, we find that retinoid potencies for the induction of hyperplasia are closely correlated to induction of overall topical irritation.

	Methods

Top Abstract Introduction Methods Results Discussion References

Animals. Female hairless mice (6-8 weeks old) were obtained from the Charles River Laboratories and maintained in an environment of controlled temperature and humidity at Allergan. Mice had access to Purina Chow and reverse-osmosis water ad libitum; animals were housed in the Allergan vivarium for 1 week before use. Protocols for animal treatment were approved by the Allergan Animal Use and Care Committee.

Materials. All-trans RA was obtained from Sigma Chemical (St. Louis, MO). Other retinoids (table 1) were synthesized at Allergan. AGN 190168 (tazarotene), AGN 190299 (the free acid of tazarotene), AGN 190121 and TTNPB [(E)-4-[2-(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)propen-1-yl]benzoic acid] are specific for the RAR but not the RXR family of receptors on the basis of both binding and transactivation (Beard et al., 1995; Nagpal et al., 1995). AGN 190727 is the meta-carboxy analog of AGN 190121 and is not active in inhibition of ODC activity (Oda et al., 1996) or in studies of receptor transactivation (data not shown). AGN 193109 is an RAR antagonist (Johnson et al., 1995). AGN 191701 [(E)-2-[2-(5,6,7,8-tetrahydro-3,5,5,8,8-pentamethyl-2-naphthyl)propen-1-yl]-4-thiophenecarboxylic acid], an analog of TTNPB, binds and transactivates the RXR (ED₅₀ = 20 nM) but appears inactive at the RARs (ED₅₀ > 10 µM) (Beard et al., 1995). SR11237 and SR11217 (Lehmann et al., 1992) are also RXR-selective retinoids with minimal activity in assays of RAR function (Nagpal et al., 1996). ¹⁴C-Ornithine was obtained from Amersham and TPA from Chemsyn (Lenexa, KS).

Hyperplasia. Mice were randomized by weight, and 3 to 6 mice were included in each treatment group. Retinoids were applied daily for 3 days to the dorsal skin of hairless mice in 100 µl of acetone, and animals were killed 72 hr after the last dose of retinoid unless otherwise noted. Epidermal thickness was measured on hematoxylin and eosin-stained paraffin-embedded sections from three separate sites on mouse dorsal skin in a blinded fashion. An ED₅₀ value for each retinoid was estimated by comparison with maximal hyperplasia induced by a retinoid active at the RARs that was run in the same experiment (eg, all-trans RA in table 3 or TTNPB in fig. 4). In some of the experiments, dosing of RA was performed under dim light, and mice were maintained in darkness for the subsequent 10 to 12 hr to prevent light-induced photoisomerization of all-trans RA. Statistical significance was determined using analysis of variance and Dunnett's test for multiple comparisons or the paired Student's t test as appropriate.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Epidermal hyperplasia in response to an RAR-specific (TTNPB) or an RXR-selective (AGN 191701) retinoid, relative to acetone vehicle-treated controls. Each point represents mean ± S.E.M. for three mice.

ODC inhibition. ODC inhibition by retinoids was determined in TPA-treated female hairless mice. Retinoids were applied dorsally in 100 µl of acetone 1 hr before TPA treatment (40 nmol/mouse) in 100 µl of acetone. Animals were killed 4 hr later, and epidermis was scraped from the dermis after treatment at 55°C for 30 sec. The supernatant of epidermal homogenate (30 min at 20,000 × g) was stored at 70°C, and ODC was assayed essentially as previously described (Gendimenico et al., 1990) except that benzethonium hydroxide was used as a trapping agent for released ¹⁴CO₂. Protein concentration in the epidermal extracts was determined by the Coomassie blue binding method (BioRad, Oakland, CA), and ODC activity was normalized to total epidermal protein content.

Topical irritation. Topical irritation of hairless mice was evaluated over a 12-day period with retinoid applied (in 100 µl of acetone) on days 1 to 5 and 8 to 11. Flaking was scored on a scale of 1 to 5 according to both severity and extent of body surface covered. For dose-response studies of total cutaneous irritation, mice were treated daily for 5 days with retinoid, and both epidermal flaking and cutaneous abrasion were scored on days 1 to 8. Dosing volume was corrected for body weight (100 µl/25 g). The maximal severity, average severity and time of onset of flaking and abrasion were used to calculate a cutaneous toxicity score (Standeven et al., 1996).

	Results

Top Abstract Introduction Methods Results Discussion References

Hyperplasia induced by RAR-specific retinoids. Retinoid induction of epidermal hyperplasia by TTNPB has been reported to require 3 to 5 days for maximal effect (Connor et al., 1986). In initial studies, we observed that all-trans RA and tazarotene, applied daily for 3 days, induced hyperplasia that was stable for an additional 3 days after cessation of treatment, with a significant decline in epidermal thickness observed on the fifth day (table 2). Retinoid potencies for epidermal hyperplasia in dose-response studies were therefore compared at 72 hr after the last of three daily treatments. Figure 1 compares all-trans RA and tazarotene, which have ED₅₀ values of ~20 and ~100 nmol/mouse, respectively.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Induction of hairless mouse hyperplasia by tazarotene and all-trans RA relative to vehicle-treated controls. Each point represents mean ± S.E.M. for four mice. Dose is daily application of retinoid per animal in nmol.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Induction of epidermal hyperplasia; shown is a comparison of AGN 190121 and a structurally related inactive retinoid analog, AGN 190727. Each point represents mean ± S.E.M. for three mice.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Effect of retinoid antagonist, AGN 193109, on epidermal hyperplasia induced by TTNPB. Hairless mice were treated topically with 3 nmol of TTNPB alone or in combination with 3 or 30 nmol of AGN 193109 and compared with acetone vehicle or AGN 193109 alone. Each observation represents mean ± S.D. for four mice. TTNPB-induced hyperplasia was inhibited by all doses of AGN 193109 tested (*, P < .05, in comparison with TTNPB treatment alone)

Action of RXR-selective retinoids. The RXR-selective compound AGN 191701 does not inhibit TPA-induced ODC activity in epidermis of the hairless mouse (Beard et al., 1995). We also tested SR 11217 and SR 11237, two additional RXR-specific synthetic retinoids (Lehmann et al., 1992; Nagpal et al., 1996), for inhibition of TPA-induced ODC activity. At a dose of 300 nmol/mouse, neither compound produced >20% inhibition of TPA-induced ODC activity. In contrast, all-trans RA, tazarotene and AGN 190121 cause 80% inhibition of TPA-induced ODC activity at doses of ~1 nmol/mouse (Nagpal et al., 1995; Oda et al., 1996). Induction of hyperplasia by an RXR-selective compound, AGN 191701, was then compared with the structurally similar RAR-specific molecule, TTNPB (fig. 4). This analysis revealed that the RAR-specific compound was substantially more potent than AGN 191701, by a factor of 1000, but that AGN 191701 exhibited some activity at the highest concentration tested (1000 nmol/mouse). The approximate ED₅₀ value of TTNPB is 0.2 nmol/mouse, whereas that of AGN 191701 is ~300 nmol/mouse. We also tested the RXR-specific compounds SR 11217 and SR 11237 in the hyperplasia assay. SR 11237 produced statistically significant induction of hyperplasia at both 100 and 1000 nmol/mouse. The magnitude of the effect of SR 11237 was greater than that of SR 11217 and showed a clear dose response (table 3). Both compounds were much less active than all-trans RA, tested simultaneously as a positive control in the same experiment. All-trans RA at 100 nmol increased epidermal thickness by 156% compared with a 73% increase for SR 11237 at 1000 nmol (table 3).

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Skin flaking in response to application of active, inactive and RXR-specific retinoids tested at a dose of 1000 nmol/animal. Each point represents the mean ± S.E.M. flaking of four or more animals. No flaking was observed in animals treated with acetone vehicle alone.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Composite topical irritation (flaking plus abrasion) in hairless mice treated for 5 consecutive days with indicated concentrations of retinoid. Each point represents mean ± S.D. of five animals.

	Discussion

Top Abstract Introduction Methods Results Discussion References

Based on similarities to the action of SDS and other irritants in skin, the possibility has been raised that epidermal hyperplasia induced by all-trans RA may be the result of nonspecific effects that are independent of receptor binding (Fisher et al., 1991; Lützow-Holm et al., 1995; Marks et al., 1990). Physical disruption of epidermal barrier function can explain the effect of nonspecific irritants in inducing epidermal hyperplasia (Proksch et al., 1991). The potential of retinoids for nonspecific irritant activity has been approximated as potency for cytolytic or detergent-like action in CHO cells in serum-free medium (Oda et al., 1996). Our data very clearly indicate that retinoid cell culture cytotoxicity is not correlated with epidermal hyperplasia. For example, the inactive retinoid AGN 190727, which has an equal cytotoxic effect as its active congener AGN 190121 in Chinese hamster ovary cells (Oda et al., 1996), does not induce hyperplasia at doses of 1000 nmol/mouse. Among active retinoids, tazarotene and AGN 190299, the free acid of tazarotene, have >20-fold reduced IC₅₀ values for CHO cell cytotoxicity compared with all-trans RA and AGN 190121, presumably by virtue of heteroatom (S, N) substitutions in the structures, but all these compounds have comparable potencies for induction of hyperplasia (table 4). Finally, TTNPB is more potent than all-trans RA in induction of hyperplasia by a factor of 100-fold but is not associated with a significant increase in cytotoxicity.

The inhibition of TTNPB-induced hyperplasia by a specific and potent RAR antagonist, AGN 193109, shows that the RAR family of receptors is a critical target for retinoid-induced hyperplasia. The binding of AGN 193109 to each of the RARs (,  and ) is approximately equal and of high affinity, in the range of 10 nM (Johnson et al., 1995). It should be noted that relative potencies for induction of hyperplasia among RAR-specific retinoids do not correlate directly with relative potencies in regulation of gene expression through the individual RARs transfected in cell culture. The potency of all-trans RA for induction of gene expression at each of the RARs is greater than or equal to TTNPB, and binding affinities of the two retinoids to the RARs are similar (Agarwal et al., 1996; Beard et al., 1995). On the other hand, TTNPB is far more potent than all-trans RA in induction of hyperplasia (table 4 and see Connor et al., 1986). Because the in vivo stability of TTNPB is significantly greater than all-trans RA (Howard et al., 1989), retinoid bioavailability also appears to be a major determinant of differences in retinoid potency in hyperplasia in addition to differences in receptor specificity. Consistent with a substantially greater stability or bioavailability for TTNPB, Connor et al. (1986) found a peak of TTNPB-induced hyperplasia in hairless mice at 5 days after a single dose of this retinoid. In contrast, we find that induction of hyperplasia by all-trans RA and tazarotene is stable for 3 days but is substantially decreased at 5 days after the cessation of treatment (table 2).

Effects of RXR-selective retinoids. Although heterodimerization with an RXR is critical for RAR transcriptional activity (Mangelsdorf et al., 1994), compounds selective for binding the RXRs induce hyperplasia only very weakly. AGN 191701 is >1000-fold less potent in induction of epidermal hyperplasia than TTNPB, a structurally related RAR-specific compound. The greatly reduced potency of AGN 191701 with respect to induction of hyperplasia most likely reflects its pharmacological properties (i.e., selectivity for the RXR family). Two additional RXR-selective retinoids, SR 11237 and SR 11217, also have an ED₅₀ value for induction of hyperplasia of >1000 nmol and appear to be >50-fold less potent than all-trans RA. SR 11237, the more active of the two, also has a higher affinity for RXR (Nagpal et al., 1996), the RXR isoform found in skin. In contrast to all-trans RA and other RAR-specific retinoids, SR 11237 is completely inactive in a separate assay of retinoid in vivo effects, reduction in the diameter of the abnormal hair follicle remnants or utriculi of the rhino mouse, at a topical dose of ~250 nmol/animal (Gendimenico et al., 1994). These data show that RXR-selective compounds are quite distinct from RAR-specific retinoids in their effects on skin pharmacology. Our data support current receptor models in which RXR-selective ligands neither activate the RAR-RXR heterodimer nor bind the RXR component when associated in the heterodimer (Forman et al., 1995; Kurokawa et al., 1994). The limited hyperplasia induced by the RXR-selective compounds can, in principle, be explained by weak interactions with the RARs. However, none of these compounds is active in inhibition of TPA-induced ODC activity, a property common to RAR-specific retinoids (table 4). Liganded RXR can, however, activate gene expression as part of an RXR homodimer or as an RXR heterodimer in association with other receptors. Although RXR homodimers are not detectable in epidermal homogenates (Fisher et al., 1994) and are unlikely to form and have activity unless highly expressed (Mangelsdorf et al., 1991; Nakshatri and Chambon, 1994; Xiao et al., 1995), liganded RXR may regulate gene expression with other heterodimeric partners (Forman et al., 1995). For example, RXR-selective ligands induce the vitamin D₃ responsive gene hydroxyvitamin D₃-24-hydroxylase in kidney (Allegretto et al., 1995). Because 1,25-dihydroxyvitamin D₃ and its analogs cause epidermal hyperplasia (Kobayasi et al., 1995), it is also possible that liganding of RXR in association with the vitamin D₃ receptor induces the weak epidermal hyperplasia observed in this study.

Epidermal hyperplasia and topical irritation. Retinoid-induced epidermal hyperplasia in the hairless mouse is well correlated with topical irritation: RXR-selective compounds are very weak in both assays, whereas relative dose responses for irritation and hyperplasia among RAR-specific compounds are very similar (compare table 4 and fig. 6). For example, the ED₅₀ value for TTNPB in induction of hyperplasia (0.2 nmol) is ~100-fold lower than all-trans RA, whereas the cutaneous toxicity dose response for TTNPB approximately parallels that for all-trans RA but is shifted again by a factor of ~100-fold to lower doses. AGN 190121 is ~2- to 5-fold less potent than all-trans RA in both assays. Finally, retinoid-induced topical toxicity, like epidermal hyperplasia, is blocked by the RAR receptor antagonist, AGN 193109 (see fig. 3 and Standeven et al., 1996). The retinoid doses at which irritation is readily apparent (composite irritation scores >5) correspond to those at which maximal or near-maximal hyperplasia is observed (e.g., 1 nmol for TTNPB and 50-200 nmol for all-trans RA and AGN 190121).