Carvacrol, a component of thyme oil, activates PPARα and γ and suppresses COX-2 expression[S]

Cyclooxygenase-2 (COX-2), the rate-limiting enzyme in prostaglandin biosynthesis, plays a key role in inflammation and circulatory homeostasis. Peroxisome proliferator-activated receptors (PPARs) are ligand-dependent transcription factors belonging to the nuclear receptor superfamily and are involved in the control of COX-2 expression, and vice versa. Here, we show that COX-2 promoter activity was suppressed by essential oils derived from thyme, clove, rose, eucalyptus, fennel, and bergamot in cell-based transfection assays using bovine arterial endothelial cells. Moreover, from thyme oil, we identified carvacrol as a major component of the suppressor of COX-2 expression and an activator of PPARα and γ. PPARγ-dependent suppression of COX-2 promoter activity was observed in response to carvacrol treatment. In human macrophage-like U937 cells, carvacrol suppressed lipopolysaccharide-induced COX-2 mRNA and protein expression, suggesting that carvacrol regulates COX-2 expression through its agonistic effect on PPARγ. These results may be important in understanding the antiinflammatory and antilifestyle-related disease properties of carvacrol.

according to the manufacturer's protocol. Chemiluminescence was captured using a cooled CCD Light-Capture camera system (ATTO, Japan) and analyzed using CS Analyzer software version 2.0 (ATTO Densitograph Software, Tokyo, Japan).

Analysis of thyme oil by GC
The thyme oil was analyzed by GC using a Hewlett Packard 5890 series II plus GC system (now Agilent Technologies, Wilmingto n, DE). A DB-17 GC column (30 m length × 0.32 mm inner diameter, 0.25 m fi lm thickness; J and W Scientifi c, Folsom, CA) was used with He as a carrier gas at a fl ow of 1 ml/min. Thyme oil (1 ml) was dissolved in ethyl alcohol (10 ml), and 1 l of this solution was injected. The split fl ow was adjusted at 50 ml/ min. The oven temperature was kept at 50°C for 2 min and increas ed to 280°C at a rate of 5°C/min. The injector temperature was 240°C. The percentage compositions were obtained from electronic integration measurements, using fl ame ionization detection (FID; 240°C). The relative FID-area percentages of the characterized components are shown in Fig. 5 .

Analysis of thyme oil by GC-MS
Thyme oil was also analyzed by high-resolution MS using a JMS BU-20 high-resolution mass spectrometer (JEOL, Tokyo, Japan) attached to a Hewlett Packard 5890 series II plus GC system. The GC conditions were the same as above. Mass spectra were taken at 70 eV under positive electron impact ionization. The mass range was m/z 50-500. The components of thyme oil were identifi ed by mass spectrum library searches conducted using the NIST/EPA/NIH Mass Spectral Search Program, Windows version 1.50, distributed by the Standard Data Reference Program of the US National Institute for Standards and Technology.
The suppression of COX-2 expression and the agonistic activity for PPARs are useful in evaluating the function of food-related components in lifestyle-related diseases. Similar to the case of 15d-PGJ 2 , it is possible that COX-2 expression is regulated by PPAR agonism exerted by chemical components in food. In this context, we evaluated commercially available lipids derived from various plants and found that six essential oils have COX-2 suppressive activity. Moreover, we identifi ed carvacrol, from the essential oil of thyme, as a major suppressor of COX-2 expression and activator of PPAR ␣ . In addition, PPAR ␥ -dependent suppression of COX-2 promoter activity was observed in response to carvacrol, indicating that carvacrol regulates COX-2 expression through its agonistic effect on PPAR ␥ .

Materials
Bergamot, castor, clove, croton, eucalyptus, fennel, linseed, olive, orange, sesame, soybean, turpentine oils, and chemical components of thyme oils were from Wako Chemicals (Japan). Cottonseed and saffl ower oils were from MP Biomedicals. Lavender, lemon, and palm oils were from Spectrum Chemicals. Thyme and rose oils were from ICN Biochemicals and Indofi ne Chemicals, respectively.

Cell culture
U937 cells ( 24 ) and BAEC ( 25 ) were grown in RPMI 1640 and DMEM, respectively, supplemented with 10% fetal calf serum and 50 M 2-mercaptoethanol. For differentiation into monocytes/macrophages, U937 cells were treated with 100 nM 12-Otetradecanoylphorbol-13-acetate (TPA) and allowed to adhere for 48 h, after which they were fed with TPA-free medium and cultured for 24 h before use ( 26 ).

Western blot analysis
Cell lysates were subjected to SDS-PAGE in 10% gels. The separated proteins were electroblotted onto polyvinylidene difl uoride membrane (Millipore), probed with human COX-2 antisera (IBL, Gunma, Japan) or anti-actin antisera (Santa Cruz), and visualized using an ECL Western blot analysis system (GE Healthcare) moter activity ( Fig. 1 ). Figure 3A shows that thyme, rose, clove, and bergamot oils had PPAR ␣ agonistic activity and that thyme oil also had PPAR ␥ agonistic activity. In the positive control experiment, selective activators of PPARs, such as Wy-14643 ( ␣ ), GW501516 ( ␤ / ␦ ), and pioglitazone ( ␥ ), had PPAR ␣ , ␤ / ␦ , and ␥ agonistic activities, respectively ( Fig. 3B ). Remarkably, thyme oil activated PPAR ␣ and ␥ with about the same dose dependency ( Fig. 4 ), within the range for suppression of the LPS-induced COX-2 promoter activity was observed ( Fig. 2A ). As a control experiment, pioglitazone suppressed LPS-induced COX-2 promoter activity in BAEC in the presence of the vector encoding PPAR ␥ (supplementary Fig. I ). In contrast, we previously found that LPS rapidly downregulates PPAR ␥ expression in U937 cells ( 12 ). By this mechanism, the suppressive effect of the PPAR ␥ agonist 15d-PGJ2 on COX-2 expression is less potent than that of dexamethasone, an agonist of the glucocorticoid receptor, which is upregulated by LPS ( 12 ). We also examined the effects of LPS on the expression of PPARs and COX-2 in U937 cells and found that PPAR ␥ mRNA alone was rapidly suppressed in response to LPS (supplementary Fig. II ).

Activation of PPARs by essential oils
A natural ligand for PPAR ␥ , 15d-PGJ 2 , suppressed the LPS-induced COX-2 mRNA expression in macrophagelike differentiated U937 cells ( 12 ). Resveratrol suppressed COX-2 mRNA expression ( 17 ) and also activated PPAR ␣ and ␥ in our cell-based assay using BAEC ( 18 ). Thus, using this assay system, we evaluated the activation of PPARs by the essential oils that suppressed LPS-induced COX-2 pro- Fig. 1. Effects on COX-2 promoter activity by essential oils in BAEC. BAEC were transiently transfected with phPES2 ( Ϫ 327/+59) together with pCMX-hPPARg1 and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 5 h with no stimulant (control) or with each oil (0.01%) and LPS (1 g/ml). The cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard and presented as the mean ± SD. *, P < 0.05, **, P < 0.01, compared with cells treated with 1 g/ ml LPS, by an unpaired t -test (n = 3).

Fig. 2.
Dose-dependent suppression of COX-2 promoter activity by thyme oil and the effects of chelerythrine on promoter activity. BAEC were transiently transfected with phPES2 ( Ϫ 327/+59), together with pCMX-hPPARg1, and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 5 h with thyme oil (0.002, 0.004, and 0.008%) and LPS (1 g/ml) (A). In a second experiment, following transfection, the cells were incubated for 5 h with no stimulant (control) or were treated with LPS (1 g/ml) in the presence or absence with chelerythrine (B). In both cases, the cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard and presented as the mean ± SD. *, P < 0.05, **, P < 0.01, compared with cells treated with 1 g/ml LPS, by an unpaired t -test (n = 3).
COX-2 promoter activity was suppressed to 48% by 0.008% thyme oil, which contained about 300 M carvacrol and 3 M borneol ( Fig. 5 ). Figure 7 shows the dose-dependent suppression of COX-2 promoter activity by carvacrol (open columns: 40% and 60% suppression by 200 and 400 M carvacrol, respectively), indicating that carvacrol is the major component of thyme oil involved in the suppression of the LPS-induced COX-2 promoter activity. On the other hand, COX-2 promoter activity was only marginally suppressed in the absence of PPAR ␥ expression vector (closed column); however, this suppression was statistically significant in the presence of the PPAR ␥ expression vector (open column) in BAEC ( Fig. 7 ), indicating that carvacrol-mediated suppression was partly PPAR ␥ dependent. In the control experiments, carvacrol alone had no effect on COX-2 promoter activity in the transfection assay using BAEC and no suppressive effect on COX-2 mRNA in U937 cells (supplementary Table II). Next, we evaluated carvacrol-mediated PPAR activation. Carvacrol activated PPAR ␣ ( Fig. 8A ) and ␥ ( Fig. 8B ) at the same concentrations at which the suppression of COX-2 promoter activity was observed ( Fig.  7 ). These carvacrol concentrations are similar to those at which PPAR ␣ and ␥ activities were observed in response to thyme oil ( Fig. 4 ). Moreover, 400 M carvacrol induced the mRNA expression of the PPAR ␣ -dependent carnitine palmitoyltransferase I gene in U937 cells (supplementary and borneol (0.60%; Fig. 5 ). Next, commercially available compounds were examined for their effects on LPS-induced COX-2 promoter activity at a concentration of 1 mM; carvacrol and borneol suppressed promoter activity by 80% and 55%, respectively ( Fig. 6 ). As shown in Fig. 2A ,   Fig. 3. Effect on activation of PPARs by essential oils. BAEC were transiently transfected with PPRE-luc together with GS-hPPAR ␣ (shaded column), pCMX-NUC1 (open column), or pCMX-hPPARg1 (solid column) and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 24 h with each essential oil (0.01%) (A). In the positive control experiment, the indicated concentration of selective PPAR activators such as Wy-14643 ( ␣ ), GW501516 ( ␤ / ␦ ), and pioglitazone ( ␥ ) were applied to the BAEC transiently transfected with PPRE-luc together with GS-hPPAR ␣ , pCMX-NUC1, or pCMX-hPPARg1, respectively (B). In both experiments, the cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard and presented as the mean ± SD. *, P < 0.05, **, P < 0.01, compared with cells treated with ethanol (control), by an unpaired t -test (n = 3).

Fig. 4.
Dose-dependent activation of PPAR ␣ and PPAR ␥ by thyme oil. BAEC were transiently transfected with PPRE-luc together with GS-hPPAR ␣ (A) or pCMX-hPPARg1 (B) and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 24 h with thyme oil (0.002, 0.004, and 0.008%). The cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard and presented as the mean ± SD. *, P < 0.05, **, P < 0.01, compared with cells treated with ethanol (control), by an unpaired t -test (n = 3).
induced COX-2 protein expression ( Fig. 10 ) is consistent with the results of the COX-2 reporter assays ( Fig. 7 ). The LPS-induced expression of COX-2 mRNA was suppressed to about 14% by treatment with 1000 M carvacrol but was not suppressed in response to 400 M carvacrol ( Fig. 9 ). This discrepancy may be due, at least in part, to the dual regulation of COX-2 mRNA at the transcriptional and posttranscriptional levels ( 30,31 ). The 3 ′ -untranslated region of COX-2 mRNA has 17 copies of the Shaw-Kamen sequence (AUUUA), which is involv ed in the regulated stabilization of COX-2 mRNA by LPS ( 31 ). Regarding this point, we found no evidence that carvacrol is involved in the control of COX-2 mRNA stability, based on reporter assays using the 3 ′ -untranslated region of COX-2 (supplementary Fig. V ). From our data, carvacrol suppressed not only promoter activity in BAEC transfected with hPCMX-hPPARg1 but also LPS-induced COX-2 mRNA and protein expression in U937 cells expressi ng PPAR ␥ .   Fig. III ). These results indicate that carvacrol is both a suppressor of COX-2 promoter activity and an activator of PPAR ␣ and ␥ .
Next, we examined whether carvacrol suppressed COX-2 mRNA and protein expression in differentiated macrophage-like U937 cells. LPS-induced COX-2 mRNA and protein expression was suppressed by carvacrol to about 14% (1,000 M in Fig. 9 ) and 40% (400 M in Fig.  10 ), respectively. Carvacrol-mediated suppression of LPS-  Fig. 6. Effects on COX-2 promoter activity by components of thyme oil and thymol. BAEC were transiently transfected with phPES2 ( Ϫ 327/+59) together with pCMX-hPPARg1 and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 5 h with no stimulant (control), with LPS alone (1 g/ml), or with LPS (1 g/ml) and each component (1 mM). The cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard. *, P < 0.05, **, P < 0.01, compared with cells treated with 1 g/ml LPS alone, by an unpaired t -test (n = 3).  7. PPAR ␥ -dependent suppression of COX-2 promoter activity in response to carvacrol. BAEC were transiently transfected with phPES2 ( Ϫ 327/+59), along with either the human PPAR ␥ expression vector, pCMX-hPPARg1 (open column), or the expression vector pCDNA3.1-GS alone (closed column) and pSV-␤ -gal as an internal control for the transfection. Following transfection, the cells were incubated for 5 h with no stimulant (control) or with carvacrol (100, 200, or 400 M) and LPS (1 g/ml). The cells were harvested, lysed, and assayed for both luciferase and ␤ -galactosidase activities. The results are represented as relative luciferase activities, which were normalized against the ␤ -galactosidase standard. *, P < 0.05, compared with open and closed columns, by an unpaired t -test (n = 3). previous fi nding that 15d-PGJ 2 , a potent natural ligand of PPAR ␥ , suppressed LPS-induced COX-2 expression in differentiated macrophage-like U937 cells and that the expression of COX-2 was regulated by a negative feedback loop mediated through PPAR ␥ ( 12 ). Taken together, it is likely that carvacrol regulates COX-2 expression, at least in part, via its agonistic effect on PPAR ␥ .

DISCUSSION
The prevention of lifestyle-related diseases, such as cardiovascular disease, diabetes, and stroke, is of worldwide interest. There have been many studies on the functionality of natural chemicals in food and drink. In this context, resveratrol in red wine is one of the most attractive of the characterized compounds, and experimental information regarding its actions has accumulated ( 32,33 ). We have focused on COX-2 ( 17 ) and PPARs ( 18 ) as possible molecular targets of resveratrol in preventing lifestyle-related diseases, while others have studied sirtuins and PGC-1 ␣ ( 34,35 ). Given that humulon ( 21,23 ) and apigenin ( 22 ), as well as resveratrol, suppress COX-2 expression and activate PPARs, we screened commercially available oils, including essential oils, for their effects on COX-2 and PPARs and found that carvacrol, a major component of thyme oil, was active. It is noteworthy that carvacrol activates PPAR ␣ and ␥ ( Fig. 8A, B ) and suppresses LPS-induced COX-2 protein expression ( Fig. 10 ) in the same concentration range. Moreover, PPAR ␥ -dependent suppression of COX-2 promoter activity was observed in response to carvacrol ( Fig. 7 ), which is consistent with our The results are represented as relative luciferase activities, which were normalized to the ␤ -galactosidase standard. *, P < 0.05, **, P < 0.01, compared with cells treated with ethanol (control), by an unpaired t -test (n = 3). Fig. 9. Suppression of COX-2 mRNA expression by carvacrol in macrophage-like differentiated U937 cells. Macrophage-like differentiated U937 cells were treated for 2 h with 10 g/ml LPS in the presence or absence of 1 mM carvacrol. Isolated RNA was used for quantitative RT-PCR analysis. The amount of COX-2 mRNA was normalized to that of GAPDH mRNA, and the normalized amount of COX-2 mRNA induced by LPS was taken as 100%. The results represent the means ± SD of three separate dishes. **, P < 0.01, compared with cells treated with 10 g/ml LPS, by an unpaired t -test (n = 3). Fig. 10. Suppression of COX-2 protein expression in response to carvacrol in macrophage-like differentiated U937 cells. Macrophage-like differentiated U937 cells were treated for 5 h with 10 g/ml LPS in the presence or absence of the indicated concentrations of carvacrol. Cellular lysate proteins (5 g/lane) were separated by 10% SDS-PAGE and transferred onto a polyvinylidene difl uoride membrane. Immunoblots were probed with antibodies specifi c for COX-2 and actin. The expression level of COX-2 was normalized against that of actin, and the normalized amount of COX-2 protein induced by LPS was taken as 1.0. The results represent the mean ± SD of three separate dishes. *, P < 0.05, **, P < 0.01, compared with cells treated with 10 g/ml LPS, by an unpaired t -test (n = 3). described effects, which include antimicrobial, antitumor, antimutagenic, antigenotoxic, analgesic, antiinfl ammatory, angiogenic, antiparasitic, antiplatelet, antihepatotoxic, and hepatoprotective properties ( 38 ). Our results may be important in understanding the antiinfl ammatory and antilifestyle-related disease properties of carvacrol, although further studies in vivo will be necessary to address the physiological signifi cance of our fi ndings. Our screening system using reporter assays for COX-2 and PPARs may be applicable for studying the functionality of food-related materials.
Because of the high concentration of carvacrol necessary to suppress COX-2 promoter activity and activate PPARs, we must consider the possible existence of a minor contaminant associated with carvacrol. To address this point, we confi rmed our results using another source of carvacrol (100% purity) via both HPLC and NMR analyses; these analyses indicated that a minor contaminant was unlikely. Another point to consider, given the high carvacrol concentration, is the possibility of nonspecifi c effects on the cell. Concerning the binding pocket of PPARs, the molecular size of carvacrol is smaller than that of other synthetic PPAR agonists, which may be one of the reasons for its relatively weak activation of PPARs. We also observed weak activation of PPAR ␤ / ␦ with 400 M carvacrol and thymol, both of which elicited an approximate 2-fold increase in activity (supplementary Fig. IV ). In contrast, carvacrol and thymol elicited 4-and 8-fold increases in PPAR ␣ activity, respectively, and 2.5-and 3.5-fold increases in PPAR ␥ activity, suggesting that carvacrol and thymol activate these factors in a somewhat specifi c manner. Fatty acids are considered to be intrinsic ligands for PPARs at higher concentrations compared with synthetic ligands. In this case, the presence of fatty acid-binding molecules, which include specifi c binding proteins and nonspecifi c interacting lipids, both inside and outside of the cells appears to modulate PPAR activation. A similar mechanism may be at work in the activation of PPARs by carvacrol and other chemical components in essential oils. Recently, carvacrol was reported to be an agonist of transient receptor potential (TRP) V3, a thermosensitive ion channel expressed predominantly in the skin and neural tissues ( 36,37 ). The EC 50 of carvacrol for TRPV3 is 0.49 mM, similar to the effective concentration for COX-2 and PPARs in our study. Moreover, thymol, but not p -cymene, is also an agonist of TRPV3 (EC 50 = 0.86 mM), suggesting a structural requirement for a hydroxyl group ( Fig. 5 ), which is also similar to our results on the effects on COX-2 and PPARs ( Figs. 6 and 8 ). Members of TRPV activate upon heating: TRPV1 at 42°C, TRPV2 at 52°C, TRPV3 at 39°C, and TRPV4 at 27 Ϫ 42°C. Further study on the relationships among COX-2, PPARs, and TRPV3 may be helpful.
We demonstrated that LPS-induced COX-2 promoter activity was suppressed by not only thyme oil but also clove, rose, eucalyptus, fennel, and bergamot oils ( Fig. 1 ). It is noteworthy that statistically signifi cant activations of PPARs were not found with rose or eucalyptus oils, but there was a tendency for PPAR ␤ / ␦ activation by thyme, clove, rose, eucalyptus, and fennel oils ( Fig. 3A ). These results may indicate that some natural chemicals in essential oils act as COX-2 suppressors but not PPAR activators, or that some minor oil components act as PPAR ␤ / ␦ agonists. Further identifi cation of the chemical compounds in these essential oils is under investigation.
In summary, we identifi ed carvacrol, a chemical component of thyme oil, as a suppressor of COX-2 and activator of PPAR ␣ and ␥ . Although most essential oils such as thyme oil have a wide range of applications in pharmaceutical, medical, and cosmetic fi elds, their mechanisms of action are not completely understood for many of the