Farnesol Stimulates Differentiation in Epidermal Keratinocytes via PPAR a *

The isoprenoids farnesol and juvenile hormone III (JH), metabolites of the cholesterol biosynthetic pathway, have been shown to stimulate fetal epidermal development in rodents. In this study we determined whether this effect might be attributed to a direct induction of keratinocytes differentiation and examined the mechanisms responsible for these effects. Rates of cornified envelope formation, a marker of keratinocyte terminal differentiation, as well as protein and mRNA levels of two proteins required for cornified envelope formation, involucrin (INV) and transglutaminase, increased 2- to 3-fold in normal human keratinocytes (NHK) treated with either farnesol or JH, even at low calcium concentrations (0.03 m M ), which otherwise in- hibit differentiation. In contrast, neither cholesterol nor mevalonate affected INV or transglutaminase mRNA levels. Effects of farnesol and JH on INV and transglutaminase mRNA levels were additive with high calcium concentrations (1.2 m M ) that independently stimulate keratinocyte differentiation. In contrast, keratinocyte DNA synthesis was inhibited by revealed with 5-bromo-4-chloro-3-indolyl phosphate/tetranitrotetrazo- lium blue substrate (Chemicon, Temecula CA), containing 2 m M levami-sole (Sigma). Hybridization with DIG-labeled sense control probes re- sulted in no signal, indicating the specificity of hybridization with the antisense probe. Omitting the DIG-labeled antisense probes from the hybridization mixture resulted in no signal, which demonstrated that only DIG-containing RNA hybrids were detected. Moreover, incubation with the 5-bromo-4-chloro-3-indolyl phosphate/tetranitrotetrazolium blue substrate reagents alone resulted in no signal, showing that en- dogenous alkaline phosphatase activity within the tissues did not contribute to the signal obtained. Statistics— Statistical analysis was performed using Student’s t test.

tion, in wild-type but not in PPAR␣؊/؊ murine epidermis. These findings suggest a novel role for selected isoprenoid cholesterol intermediates in the regulation of differentiation-specific gene transcription and a convergence of PPAR␣ with the cholesterol synthetic pathway.
Mammalian epidermis is comprised of stratifying, progressively differentiating keratinocytes, with the cells of each strata displaying a gene expression pattern that reflects the extent of their differentiation (reviewed in Ref. 1). Although proliferative cells in the basal layer express keratin genes, such as K5 and K14, suprabasal cells sequentially express K1 and K10 (2)(3)(4), involucrin (INV) 1 (5,6), the calcium-dependent membrane-bound cross-linking enzyme transglutaminase (TG'ase), the intermediate filament-associated protein profilaggrin, loricrin, and other structural protein constituents of the cornified envelope (CE) (7,8). Many of the genes expressed in these suprabasal layers are necessary for the formation of the CE, the specialized external envelope of the terminally differentiated corneocyte, which confers rigidity and mechanical resistance (9). Corneocytes, the critical structural components of the stratum corneum, together with extracellular lipids, provide the outermost epidermal layer with a barrier to systemic water loss, necessary for life in a terrestrial environment (10,11).
Increased extracellular calcium, resulting in elevated intracellular calcium, is perhaps the best known signal for the induction of normal human keratinocyte (NHK) differentiation (12,13). Other extra-and intracellular signals by which NHK differentiation is controlled are not yet fully understood. Numerous studies suggest an important role for RXR-heterodimerizing nuclear hormone receptors in the regulation of epidermal differentiation. The most intensely studied members of this group are retinoic acid receptor (RAR) isoforms. For example, overexpression of a truncated inactive RAR␣ during fetal development results in aberrant epidermal development (14,15), and both systemic and topical RAR ligands modulate epidermal growth and differentiation (16 -18). Ligands of the vitamin D receptor also exert profound, but often opposing, effects from RAR ligands (16,19,20). Additionally, we have shown that ligands and activators of other RXR-interacting receptors, thyroid hormone receptor, peroxisome proliferatoractivated receptor (PPAR␣), and LXR, accelerate both epidermal maturation and the formation of a competent barrier in rodents (21)(22)(23). Furthermore, in normal human keratinocytes, activators of PPAR␣ and LXR, like vitamin D receptor ligands, stimulate differentiation and inhibit proliferation (24,25).
The farnesoid X-activated receptor (FXR) is another member of the RXR-interacting family. FXR, and its murine homolog RIP14, have been demonstrated in several tissues that are active in sterologenesis such as liver, gut, adrenal gland, and kidney (26). In early studies, farnesol, an isoprenoid in the mevalonate pathway, and the farnesoid metabolite, juvenile hormone III (JH), which regulates metamorphosis in insects (26), were described as activators of FXR⅐RXR complexes (26). Recently, bile acids have been identified as physiologic ligands for FXR in tissues that express a bile acid transporter such as liver and intestine (27)(28)(29). In liver, activation of FXR inhibits diversion of cholesterol toward bile acid synthesis by decreasing the expression of cholesterol 7␣-hydroxylase (27)(28)(29).
Previous studies have shown that epidermis is one of the most active sites of steroidogenesis in mammals (30). Based upon those early observations, we assessed whether farnesol and JH might regulate growth and differentiation. We showed recently that both of these isoprenoids accelerate epidermal development in fetal rats (21,23). In the present study we explore the mechanisms responsible for these effects. We show here that farnesol directly stimulates differentiation in NHK as well as in adult murine epidermis. Furthermore, these effects are not mediated by FXR, but rather by PPAR␣, suggesting that the regulation of keratinocyte differentiation by isoprenoid sterol intermediates is dependent upon PPAR␣ activation.
Keratinocyte Differentiation Markers-The rate of CE formation and protein levels of involucrin and transglutaminase were determined as described previously (24).
RNA Isolation, Northern Blotting, and cDNA Probes-Total RNA was isolated with Trizol (Sigma) following the manufacturer's protocol. Poly(A) ϩ mRNA was isolated as described previously (32). RNA (15 g per sample) or Poly(A) ϩ mRNA (8 g per sample) was size-fractionated through a 1% agarose gel containing 2.2 M formaldehyde, as described previously (24). RNA integrity was visualized following acridine orange staining of the electrophoresed gel. The RNA was transferred to a nylon membrane that was subsequently baked at 80°C for 2 h. Blots were hybridized with the appropriate 32 P-labeled probe: P1-2 for involucrin (a gift from Dr. Howard Green, Harvard University); hTG for keratinocyte transglutaminase 1 (a gift from Dr. Robert Rice, UC Davis); or PPAR␣ (a 620-bp fragment encoding the ligand binding domain, a gift from Dr. N. Bass, UCSF) overnight at 65°C. Washes were then performed in a solution containing 0.1% SSC and 0.1 SDS for 20 min at room temperature, followed by a 20-min wash at 65°C. Autoradiography was performed at Ϫ70°C. Blots were probed with ␤-actin to confirm equal loading. Bands were quantified by densitometry.
DNA Synthesis-The rate of DNA synthesis was determined as described previously (24)  RNA Detection by RT-PCR-0.8 g total RNA was reverse-transcribed with 20 ng of random hexamer (RT-for-PCR kit, CLONTECH, San Diego, CA) at 42°C for 2 h. As a control experiment, cDNA synthesis reactions were carried out without reverse transcriptase (RT) for every RNA sample to evaluate DNA contamination. The PCR mixture (50 l) contained a 0.4 mM final concentration of each (forward and reverse) primer, 0.2 mM of each deoxynucleotide triphosphate, and 1 unit of Taq DNA polymerase in 1ϫ PCR buffer (all from CLONTECH). FXR-specific cDNA products were amplified by PCR with the following oligonucleotides: forward primer: 5Ј cgt gac ttg cgn caa gtg acc 3Ј, reverse primer: 5Ј cca nga cat cag cat ctc agc g 3Ј, designed to yield a 683-bp product. Samples were heated to 97°C for 5 min, followed by 35 cycles of denaturation at 97°C for 1 min, annealing at 68°(primer set 1) for 1 min, and extension at 72°C for 1 min, with a final extension at 72°C for 5 min, using a Perkin-Elmer Thermal Cycler (Model 480). PCR products were separated by electrophoresis on a 1.5% agarose gel and visualized with ethidium bromide staining. A 1-kilobase (kb) DNA ladder (Life Technologies, Inc.) was used for the molecular weight markers. The identity of the amplified PCR product was confirmed as an FXR cDNA by Southern analysis.
Transient Transfections-NHK were transfected as described (33) with minor modifications. Briefly, primary keratinocytes were passaged onto 60-mm dishes 1 day prior to transfection to yield a confluence of 20 -40% on the day of transfection. 2 g of INV-luciferase construct (33) or PPRE-luciferase construct (a gift from Dr. N. Bass, UCSF), 0.2 g of RSV-␤-gal (a reporter plasmid used as an internal control that does not respond to the activators used), and 6.5 g of dihexabromide (Polybrene, Aldrich) was added in media (KGM containing 0.03 mM Ca 2ϩ ) in a final volume of 0.65 ml. The cells were incubated at 37°C for 5 h with gentle shaking each hour. The INV constructs used have been previously described (33). Cells were then rinsed with CMF-PBS, followed by incubation at room temperature for 3 min with 10% glycerol in media. Following two rinses with CMF-PBS, keratinocytes were incubated overnight with 2 ml of KGM containing 0.03 mM Ca 2ϩ . Keratinocytes were treated the following day for either 24 or 48 h as indicated for each experiment. Cells were rinsed twice with cold PBS and harvested in 250 l of reporter lysis buffer (Promega, Madison, WI). The lysate was spun at 10,000 ϫ g (4°C) for 2 min, and 10 -20 l of supernatant was assayed with luciferase substrate (Promega) and ␤-galactosidase substrate (Galacto-light Tropix, Bedford, MA) following the manufacturer's instructions. ␤-Galactosidase activity was used to normalize data and correct for variations in transfection efficiencies.
CV-1 cells were incubated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and transfected with 2.0 g of PPRE-luciferase, 2.0 g of PPAR␣ or pCMX, and 0.2 g of RSV-␤galactosidase. Cells were treated the following day with farnesol or vehicle and harvested and assayed as above.
In Vivo Treatment and Tissue Preparation-Shaved areas of conventional hairy mice were treated on one flank with 40 l/cm 2 1 mM farnesol (in propylene glycol/ethanol, 7:3) treated twice a day for 3 consecutive days. Control animals were treated with vehicle alone. Animals were sacrificed, the flank skin was excised, and skin samples were secured to dissecting trays with pins. The dissecting trays were then flooded with fixative, and after an initial 20-to 30-min fixation with 4% formaldehyde (prepared freshly from paraformaldehyde powder) in PBS the samples were cut into longitudinal strips to preserve the orientation of the tissue samples. The tissue pieces were transferred to vials containing the fixative, and fixation continued at 4°C for 12 h. The samples were dehydrated in ethanol and embedded in paraffin. Sections (5 m) were cut on a Leica (Deerfield, IL) RM1350 microtome and collected on Superfrost Plus microscope slides (Fisher Scientific, Pittsburgh, PA). The sections were either stained with hematoxylineosin or used for immunohistochemistry or in situ hybridization.
In Situ Hybridization-Digoxigenin-labeled RNA probes to detect loricrin and profilaggrin mRNAs were made from linearized cDNA sequences (a gift from S. Yuspa, National Institutes of Health) using reagents supplied by Roche Molecular Biochemicals. In situ hybridization was performed as described previously (22). The sections were hybridized at 40°C, and the hybridization of DIG-labeled probes to the endogenous mRNA was detected by anti-DIG-alkaline phosphatase (Roche Molecular Biochemicals). Alkaline phosphatase activity was revealed with 5-bromo-4-chloro-3-indolyl phosphate/tetranitrotetrazolium blue substrate (Chemicon, Temecula CA), containing 2 mM levamisole (Sigma). Hybridization with DIG-labeled sense control probes resulted in no signal, indicating the specificity of hybridization with the antisense probe. Omitting the DIG-labeled antisense probes from the hybridization mixture resulted in no signal, which demonstrated that only DIG-containing RNA hybrids were detected. Moreover, incubation with the 5-bromo-4-chloro-3-indolyl phosphate/tetranitrotetrazolium blue substrate reagents alone resulted in no signal, showing that endogenous alkaline phosphatase activity within the tissues did not contribute to the signal obtained.
Statistics-Statistical analysis was performed using Student's t test.

Farnesol and JH Induce Keratinocyte Differentiation and Inhibit
Proliferation-To determine whether farnesol or JH affect the differentiation of keratinocytes, we first measured the rate of cornified envelope formation, the final product of terminal differentiation, in NHK incubated either in low (0.03 mM) or high (1.2 mM) calcium in the presence of JH, farnesol, or vehicle (dimethyl sulfoxide or ethanol). As shown in Fig. 1, CE formation was increased approximately 2-fold by either farnesol or JH in cells incubated in low calcium. Incubation of NHK in 1.2 mM calcium resulted in a 4-fold increase in CE formation, and farnesol or JH treatment together with 1.2 mM calcium resulted in an 8-to 11-fold increase over low calcium controls (Fig. 1).
We next measured protein levels of INV and transglutaminase, early markers of differentiation, in NHK incubated in 0.03 mM calcium and treated with farnesol. Farnesol induced INV and transglutaminase protein levels in a dose-dependent manner, with 10 -15 M resulting in a 2-to 3-fold increase ( Fig.  2A). In contrast, protein levels were not affected by treatment with other sterol metabolites such as cholesterol or mevalonate (not shown). JH also resulted in increased INV and transglutaminase protein levels (approximately 2-fold), with maximal effects observed with 15 M (data not shown). A similar induction of INV and transglutaminase protein was observed in cells incubated in high calcium and treated with farnesol or JH (Fig.  2B).
We next used Northern blot analysis to determine whether increased mRNA levels might underlie the induction of INV and transglutaminase protein by either farnesol or JH. As shown in Fig. 3 (A and B), farnesol and JH significantly increased INV and transglutaminase mRNA levels in a dose-dependent manner. In contrast, treatment with either cholesterol or mevalonate did not affect mRNA levels (data not shown). Both JH and farnesol further increased INV and transglutaminase mRNA levels in NHK incubated in 1.2 mM calcium (Fig.   3C). These data indicate that farnesol and JH stimulate differentiation in NHK in both low and high calcium, suggesting that the mechanism by which these compounds increase differentiation differs from that of calcium.
Cells induced to differentiate either by increased extracellular calcium or by 1,25-dihydroxyvitamin D 3 or PPAR␣ activators simultaneously exhibit a decrease in cellular proliferation (19,24,34). To determine whether cell growth is similarly inhibited by farnesol, we next compared the rates of DNA synthesis in farnesol-and vehicle-treated NHK. As shown in Fig. 4, the rate of DNA synthesis was decreased approximately 85% in keratinocytes incubated in the presence of 10 M farnesol for 24 h and labeled with [ 3 H]thymidine for the final hour. These results were dose-dependent, and similar results were obtained with JH (not shown). These data indicate that farnesol and JH not only stimulate differentiation but also inhibit proliferation.
Localization of a Farnesol-responsive Region in the INV Gene-We next sought to determine whether the increase in INV or transglutaminase mRNA levels induced by farnesol might be attributable to increased transcription. Keratinocytes were transiently transfected with a 3.7-kb INV promoter or To further localize farnesol-responsive regions in the INV promoter, we next measured the effects of farnesol on activity in constructs in which truncations or internal deletions were made (Fig. 5B). Internal deletion of the intron at ϩ182 bp to ϩ1228 bp (construct designated 2.7 kb) or of the regions containing Ϫ1880 bp to Ϫ156 bp and Ϫ3 bp to ϩ1228 bp (construct designated INV-P) did not significantly affect responsiveness to farnesol. In contrast, deletion of the region containing Ϫ2452 bp to Ϫ1880 bp (construct designated MP-1) resulted in loss of responsiveness (Fig. 5B). Similar results were obtained with JH (data not shown). These studies demonstrate that the sequence spanning Ϫ2452 bp to Ϫ1880 bp of the INV promoter contains a region required for increased transcription by farnesol. We have previously shown the same region to be responsive to PPAR␣ activators (24).
An AP-1 Site (Ϫ2117 to Ϫ2111 bp) Is Essential for Increased INV Transcription by Farnesol-A functional AP-1 site, which mediates the increase in INV transcription following treatment with phorbol esters, has been identified within the region spanning Ϫ2452 bp to Ϫ1880 bp of the INV promoter (35,36). We have found that this AP-1 site is important for increased INV transcription by PPAR␣ activators. 2 To determine whether this AP-1 site is also important for the induction of INV by farnesol, reporter activity was measured in keratinocytes transfected with a construct containing a wild-type AP-1 response element (Ϫ2117 to Ϫ2111 bp) or with a corresponding construct in which the AP-1 site has been mutated (TGAGTCA mutated to TGAGCCA). As shown in Fig. 6, the inducibility of reporter activity by farnesol was abolished by the AP-1 mutation. Thus, farnesol and PPAR␣ activators stimulate INV transcription via a similar AP-1-dependent pathway.
FXR mRNA Is Not Present in NHK-In agreement with recent studies that suggest that farnesol is most likely not a physiologic activator of FXR, we did not detect FXR in NHK by Northern blot analysis (data not shown). Similarly, using RT-PCR, FXR mRNA was not found in NHK (Fig. 7, lanes K1 and  K2), whereas an intense band of the expected size was detected in preparations of HepG2 cells, a human liver cell line, used as a positive control (Fig. 7, lanes H1 and H2). In contrast, and similar to results of other laboratories (37), PPAR␣ mRNA was detected in NHK by Northern analysis (see Fig. 9). These data indicate that the effects of farnesol on NHK differentiation are not mediated by FXR.
PPRE Transactivation and PPAR␣ mRNA Levels Are Increased by Farnesol and JH in NHK-These studies suggest that farnesol, JH, and PPAR␣ activators, stimulate INV transcription via similar mechanisms. To determine whether farnesol and JH might stimulate the PPAR response element (PPRE), we next measured their effects on the activity of a PPRE reporter gene transiently transfected into keratinocytes. The PPAR activators Wy-14643 or clofibric acid induced reporter activity between 2-and 3-fold (Fig. 8A), similar to reports by other laboratories (14). Farnesol or JH also stimulated reporter activity (JH 2.1-fold; farnesol 3.4-fold) (Fig. 8A). Furthermore, concentrations of farnesol and Wy-14643, which did not maximally stimulate PPRE activity alone, exerted additive effects together, whereas maximal concentrations together did not increase activity further (Fig. 8B). Similar results were obtained with farnesol and clofibric acid (not shown). These data indicate that farnesol and JH activate a PPRE in NHK and suggest that this may be the mechanism for their induction of differentiation.
PPAR␣ activators increase PPAR␣ mRNA levels in the liver (38,39), and auto-induction of nuclear receptor mRNA levels by their activators have been shown in other tissues (40). Here, we show that PPAR␣ mRNA levels in NHK are increased by treatment with clofibric acid (Fig. 9). Moreover, farnesol also increases PPAR␣ mRNA levels (Fig. 9), suggesting that farnesol exerts its effects on keratinocyte differentiation via the PPAR␣ signaling pathway. Induction of PPRE Activity by Farnesol Requires PPAR␣-Several receptor combinations can bind to a PPRE. To determine if farnesol-induced PPRE transcription is mediated by PPAR␣, CV-1 cells, which do not contain endogenous PPARs, were cotransfected with a PPRE-luciferase construct together with expression vectors for either control pCMX or PPAR␣. Basal levels of PPRE activity were near background in the absence of PPAR␣, and farnesol had no significant effect on activity (Fig. 10). In contrast, cotransfection of PPAR␣ resulted in increased basal PPRE activity, and this activity was in- Farnesol Increases Differentiation of Normal Murine Skin-Recent studies by our laboratory have demonstrated that topical application of PPAR␣ activators stimulates differentiation in adult murine skin. 2 We next asked whether topical application of farnesol also affects epidermal differentiation in vivo. Farnesol did not produce morphological alterations in hematoxylin-eosin-stained sections of treated versus control murine skin (data not shown). However, by immunohistochemistry, farnesol increased the expression of two structural proteins, which are expressed in the spinous and granular layers, profilaggrin/filaggrin (Fig. 11: vehicle (A) versus farnesol treatment (E)), and loricrin ( Fig. 11: vehicle (C) versus farnesol treatment (G)). Furthermore, detected by in situ hybridization, mRNA levels of profilaggrin/filaggrin (Fig. 12, A versus E) and loricrin (Fig. 12, G versus C), were also markedly increased in farnesol-treated versus vehicle-treated wild-type mice. These observations indicate that topical farnesol treatment promotes terminal differentiation of the epidermis in vivo.
PPAR␣ Mediates the Stimulatory Effects of Farnesol on Epidermal Differentiation-We next determined whether farnesol exerts its effects on epidermal differentiation via PPAR␣, by examining the effects of farnesol or vehicle on mice lacking functional PPAR␣ receptors (PPAR␣Ϫ/Ϫ) versus their normal littermates (PPAR ␣ϩ/ϩ). Farnesol treatment had no effect on protein levels of filaggrin (Fig. 11F), or loricrin (Fig. 11H) by immunohistochemistry in the PPAR␣Ϫ/Ϫ mice. Additionally, by in situ hybridization, mRNA levels of filaggrin (Fig. 12F) and loricrin (Fig. 12H) in PPAR␣Ϫ/Ϫ mice were unchanged following farnesol treatment. These data indicate that the stimulatory effects of farnesol on epidermal differentiation requires the presence of PPAR␣.

DISCUSSION
Recent studies have revealed that an increasing number of endogenous lipid metabolites, such as lipid-soluble vitamins, fatty acids, and oxysterols, share properties of classic hormones; i.e. they interact directly with nuclear hormone receptors to regulate transcription. Farnesol is a 15-carbon isoprenoid derived from farnesylpyrophosphate, a key intermediate in the cholesterol biosynthetic pathway. The ability of farnesol to accelerate the development of the fetal epidermal permeability barrier led us to ask whether farnesol directly affects differentiation and proliferation in epidermal keratinocytes. Our previous work has shown that fatty acids and oxysterols, acti- vators of the nuclear hormone receptors PPAR␣ and LXR, respectively, stimulate epidermal barrier ontogenesis and keratinocyte differentiation (24,25). Taken together, these observations suggest that the epidermis, with its high rate of fatty acid and cholesterol synthesis, generates endogenous metabolites that locally regulate keratinocyte growth and differentiation. We explored here whether the sterol intermediates farnesol or JH also regulate NHK growth and differentiation. We show that treatment of NHK with either farnesol or JH results in increased cornified envelope formation and in increased mRNA and protein levels of the differentiation-specific genes INV and transglutaminase, with a parallel inhibition of DNA synthesis.
The effects of ligands and activators of nuclear receptors on keratinocyte differentiation in vitro can differ dramatically from those effects that are observed in vivo. For example, all-trans-retinoic acid, the ligand for the retinoic acid receptor, inhibits differentiation in vitro, whereas topical treatment of the intact animal does not suppress differentiation (17,18). Conversely, differentiation is stimulated in vitro but inhibited in vivo by 1,25-dihydroxyvitamin D 3 , the vitamin D receptor ligand (16,20,41). Thus, we also examined the effects of farnesol on murine epidermal differentiation in vivo. Farnesol increased the expression of loricrin and filaggrin both at the protein and mRNA level in wild-type mice, indicating that farnesol promotes differentiation in vivo as well as in vitro.
Farnesol and JH increase INV and transglutaminase promoter activity, indicating regulation of these genes at the tran-scriptional level. Our experiments with deletional INV constructs revealed a region of the INV promoter (Ϫ2452 bp to Ϫ1880 bp) that was responsive to farnesol and JH. Mutation of the AP-1 site within this region abolished farnesol responsiveness, indicating that this AP-1 site is involved in the transcriptional regulation of INV by farnesol. This site also mediates the INV transcriptional response to calcium (33). 2 Our studies suggest, however, that farnesol and JH stimulate differentiation at least in part by a pathway independent of calcium, because in the presence of maximally stimulatory calcium levels, farnesol further increased INV transcription and mRNA levels.
Several members of the steroid/nuclear receptor superfamily functionally interact with AP-1 proteins such as c-Jun and c-Fos, which are expressed in differentiated keratinocytes (42). Both positive and negative interactions have been reported, depending on cell type (43, 44 and refs. therein). The cell type specificity is determined in part by the composition of the AP-1 complexes and by interactions with additional transcriptional regulators. Future studies to determine whether nuclear receptors and AP-1 or other transcription factors present in NHK interact to regulate keratinocyte differentiation will contribute to the elucidation of the cascade of events leading to the transcriptional induction of genes such as INV.
The farnesol-responsive region on the INV gene (Ϫ2452 bp to Ϫ1880 bp) which we have localized in the present study is also responsive to PPAR␣ activators (24). Thus we explored the possibility that farnesol-regulated INV transcription occurs via PPAR␣-mediated pathways. We demonstrate here that farnesol transactivates a PPRE in NHK, and that farnesol and the PPAR␣-specific activator Wy-14643, when applied together in suboptimal doses, exert additive effects on PPRE activation, whereas no further effects are observed with both agents at maximal concentrations. These results are similar to the combined effects of these agents on fetal epidermal barrier development (21), which also suggests a common mechanism of action. Additionally, we show that farnesol, like PPAR␣ activators, regulates mRNA levels of PPAR␣ in NHK. Furthermore, in cells that do not contain endogenous PPARs, we demonstrate that PPAR␣ is activated by farnesol, similar to studies by another laboratory (45).
To clearly delineate the role of PPAR␣ in farnesol-stimulated epidermal differentiation, we examined the effects of topical farnesol on normal murine skin and in mice lacking functional PPAR␣ (PPAR␣Ϫ/Ϫ). In previous studies we found that the stimulatory effects of PPAR␣ activators on epidermal differentiation are not observed in PPAR␣Ϫ/Ϫ mice, indicating that the effects of these compounds are mediated by PPAR␣. 2 The in vivo studies presented here provide definitive evidence that farnesol also exerts its effects on keratinocyte differentiation via PPAR␣ activation, because farnesol did not stimulate the expression of profilaggrin or loricrin protein or mRNA in PPAR␣Ϫ/Ϫ mice. PPAR␣ is a promiscuous receptor that is activated by a wide variety of compounds, including fibrates and fatty acids. Direct binding of farnesol to PPAR␣ has not been observed 3 ; whether PPAR␣ may be activated by a farnesoid metabolite awaits further studies.
Fatty acids and cholesterol are major constituents of mammalian cell membranes, and their cellular levels must be regulated so as to provide a balanced supply of these lipids during normal cellular growth. Regulation of lipid synthesis is of further importance in the epidermal keratinocyte, because the generation of large quantities of cholesterol and fatty acid in the proper ratio is critical for the maintenance of the epidermal 3 Communication with T. Willson, Glaxo-Wellcome. permeability barrier (46). Our results indicate that one mechanism by which this balance may be maintained is through activation of PPAR␣, which regulates many of the genes involved in fatty acid catabolism, by at least one intermediate in the cholesterol synthetic pathway, farnesol. Another point of transcriptional regulation at which the sterol and fatty acid synthetic pathways converge is through activation of sterol regulatory element binding proteins (SREBPs) by cellular sterols (47). SREBPs are membrane-bound transcription factors which, when intracellular sterol levels decrease, are proteolytically cleaved to release the mature form (47). This truncated mature form then can enter the nucleus and regulate genes important for cholesterol and fatty acid homeostasis, such as hydroxy-methylglutaryl-coenzyme A synthase and reductase, fatty acid synthase, and acetyl-CoA carboxylase (47)(48)(49). Thus, cholesterol can regulate fatty acid and cholesterol synthesis in concert via SREBPs, whereas compounds derived from intermediates in the cholesterol synthetic pathway can regulate fatty acid metabolism via PPAR␣. This provides the cell with multiple mechanisms with which to maintain a balance between two essential membrane lipids and extends the role of isoprenoids beyond protein prenylation and stimulation of apoptosis and inhibition of DNA synthesis (50,51). Finally, previous studies have shown that PPAR␣ activators can ameliorate the epidermal hyperplasia that occurs in essential fatty acid-deficient mice and in murine skin following repeated barrier disruption, 2 suggesting that farnesol may be useful as a topical agent for the treatment of various dermatoses characterized by aberrant growth and differentiation.
In summary, the results presented here reveal a novel role for the isoprenoid farnesol in PPAR␣-regulated transcription of differentiation-specific genes and suggest a convergence of the biosynthetic pathways of two of the major lipid species in the skin, fatty acids and cholesterol.