Toxicology Letters 262 (2016) 92–99

Parabens inhibit fatty acid amide hydrolase: A potential role inparaben-enhanced 3T3-L1 adipocyte differentiation

Sean D. Kodania, Haley B. Overbyb, Christophe Morisseaua, Jiangang Chenb,c, Ling Zhaob,Bruce D. Hammocka,*aDepartment of Entomology and Nematology, and UC Davis Comprehensive Cancer Center, University of California, Davis, CA 95616, USAbDepartment of Nutrition, The University of Tennessee, Knoxville, TN 37996, USAcDepartment of Public Health, The University of Tennessee, Knoxville, TN 37996, USA

H I G H L I G H T S

� Parabens inhibit the endocannabinoid enzyme fatty acid amide hydrolase (FAAH).� Paraben inhibition has time-independent, mixed-type kinetics.� Benzylparaben, the most potent paraben, inhibits FAAH with sub-micromolar potency.� Endocannabinoids may mediate paraben-enhanced adipogenesis but not through CB1 activation.

A R T I C L E I N F O

Article history:Received 18 May 2016Received in revised form 2 August 2016Accepted 18 September 2016Available online 19 September 2016

Keywords:ParabensBenzylparabenFatty acid amide hydrolaseAdipocyte

A B S T R A C T

Parabens are a class of small molecules that are regularly used as preservatives in a variety of personalcare products. Several parabens, including butylparaben and benzylparaben, have been found to interferewith endocrine signaling and to stimulate adipocyte differentiation. We hypothesized these biologicaleffects could be due to interference with the endocannabinoid system and identified fatty acid amidehydrolase (FAAH) as the direct molecular target of parabens. FAAH inhibition by parabens yields mixed-type and time-independent kinetics. Additionally, structure activity relationships indicate FAAHinhibition is selective for the paraben class of compounds and the more hydrophobic parabens havehigher potency. Parabens enhanced 3T3-L1 adipocyte differentiation in a dose dependent fashion,different from two other FAAH inhibitors URB597 and PF622. Moreover, parabens, URB597 and PF622 allfailed to enhance AEA-induced differentiation. Furthermore, rimonabant, a cannabinoid receptor 1 (CB1)-selective antagonist, did not attenuate paraben-induced adipocyte differentiation. Thus, adipogenesismediated by parabens likely occurs through modulation of endocannabinoids, but cell differentiation isindependent of direct activation of CB1 by endocannabinoids.

ã 2016 Elsevier Ireland Ltd. All rights reserved.

Contents lists available at ScienceDirect

Toxicology Letters

journal homepage: www.elsev ier .com/locate / tox let

Abbreviations: AADAC, arylacetamide deacetylase; AEA, arachidonoyl ethano-lamide; BnP, benzylparaben; BuP, butylparaben; CB1R, cannabinoid receptor 1;CB2R, cannabinoid receptor 2; CMNA, cyano(6-methoxynaphthalen-2-yl)methylacetate; EC, endocannabinoid; FAAH, fatty acid amide hydrolase; GR, glucocorticoidreceptor; hCE1, carboxylesterase 1; hCE2, carboxylesterase 2; OMP, N-(6-methoxypyridin-3-yl) octanamide; PF-622, N-phenyl-4-(quinolin-2-ylmethyl)pi-perazine-1-carboxamide; PPARg, peroxisome proliferator-activated receptor g;URB597, 30-carbamoyl-[1,10-biphenyl]-3-yl cyclohexylcarbamate.* Corresponding author at: Department of Entomology and Nematology,

University of California, Davis, Davis, CA 95616, USA.E-mail address: bdhammock@ucdavis.edu (B.D. Hammock).

http://dx.doi.org/10.1016/j.toxlet.2016.09.0110378-4274/ã 2016 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The biologic activity of many high production volume smallmolecules in consumer care products is poorly understood but is ofhigh concern due to their daily use. Compared to environmentaltoxicants where individuals are unintentionally exposed to lowdoses of chemicals, many of these chemicals are voluntarilyapplied at high doses directly to the skin. Of these consumer careproduct chemicals, parabens are a class of small moleculescommonly added to pharmaceuticals, cosmetics and food productsfor their antimicrobial activity (Bledzka et al., 2014). Structurally,they are esters of p-hydroxybenzoic acid and most commonly

S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99 93

include methylparaben, propylparaben and butylparaben. They arenot acutely toxic with oral LD50 values > 1 g/kg (Soni et al., 2005).Furthermore, methylparaben and propylparaben are on the Foodand Drug Administration’s list of chemicals “generally recognizedas safe” (FDA, 1973). However, despite the low acute toxicity therehas been significant concern over whether they act chronically asendocrine disrupting chemicals with studies providing evidencefor (Chen et al., 2007) and against (Hoberman et al., 2008) sucharguments.

In addition to their possible endocrine disrupting effects as sexhormone mimics, parabens have been reported to stimulateadipocyte differentiation (Taxvig et al., 2012; Hu et al., 2013).Adipocytes are formed early in life and the number of adipocytesdoes not generally increase in lean adults. However, it is possiblefor precursor cells to differentiate into adipocytes in adults whenthe demand for energy storage increases (Wang et al., 2014).Paraben-enhanced differentiation would supplement this processand increase the prevalence of obesity within exposed populations.Adipogenic effects are suggested to be mediated through theperoxisome proliferator-activated receptor gamma (PPARg) andthe glucocorticoid receptor (GR). Antagonism of either PPARg orthe GR using small molecules partially reduced the paraben-enhanced differentiation (Hu et al., 2013). Butylparaben, but notbenzylparaben, activate the PPARg receptor and all parabenstested, including butylparaben and benzylparaben, activated theGR but do not compete with dexamethasone binding to the GR (Huet al., 2013; Pereira-Fernandes et al., 2013). Thus, while PPARg andGR may be involved in paraben-mediated differentiation, they arenot the direct targets responsible for all of these effects.

In addition to PPARg and the GR, evidence exists to implicatethe endocannabinoid (EC) system in the regulation of adipogenesisand general energy homeostasis (Silvestri and Di Marzo, 2013).Signaling through the EC system primarily occurs through twolipid mediators, arachidonoyl ethanolamide (also known asanandamide, AEA) and 2-arachidonoylglycerol (2-AG) that bindto the cannabinoid receptors (CB1 and CB2) (Pertwee, 2015).Independent of food intake, blockage of the CB1 receptor reduceslipogenesis (Vida et al., 2014). The CB1 antagonist rimonabant wasused in clinical trials for the treatment of obesity, but it was pulledfrom the market due to serious side effects including depression(Christopoulou and Kiortsis, 2011). Pharmacologic and geneticknockout of the CB1 receptor seems to reduce or ablateglucocorticoid stimulated weight gain, indicating these effectsoccur downstream of the GR (Bowles et al., 2015). In addition totheir action on the CB1 receptor, endocannabinoids may targetadditional pathways to affect metabolism. For example, AEA hasbeen shown to bind directly to PPARg to mediate adipocytedifferentiation (Bouaboula et al., 2005; Karaliota et al., 2009).

Given the role of the EC system on adipocyte differentiation, wehypothesized previously observed effects of parabens on adipo-cytes may be mediated through modulation of endocannabinoidsignaling. To test this hypothesis, we directly tested the ability ofparabens to inhibit fatty acid amide hydrolase (FAAH), the enzymeprimarily responsible for regulating concentrations of AEA. Inaddition, the effects of parabens, other FAAH inhibitors, and theirinteraction with AEA on 3T3-L1 adipocyte differentiation wereexplored.

2. Materials and methods

2.1. Chemicals

Methylparaben, propylparaben, butylparaben, benzylparabenand 4-hydroxybenzoic acid were all purchased from AcrosChemicals. URB597 and PF622 were purchased from CaymanChemical and heptylparaben was purchased from Sigma-Aldrich.

Other paraben-like compounds described were synthesized asdescribed in Supplementary material. N-(6-methoxypyridin-3-yl)octanamide (OMP) and cyano(6-methoxynaphthalen-2-yl)methylacetate (CMNA) were synthesized as previously described (Shanand Hammock, 2001; Huang et al., 2007).

2.2. Preparation of enzyme extracts

The transgenic production of the FAAH enzyme and otheresterases in baculovirus are previously described (Nishi et al.,2006; Huang et al., 2007). A crude preparation of enzyme wasprepared by centrifuging cells at 1000 rpm, 15 min, 4 �C and re-suspending the pellet in 50 mM tris buffer (pH = 8.0) with 1 mMbenzamidine and 1 mM EDTA. The solution was homogenized(3 � 15 s) and centrifuged (9000g, 20 min). The resulting pellet wasre-suspended in 50 mM tris/HCl buffer (pH = 8.0) with 1 mMCHAPS and 10% glycerol and kept frozen at �80 �C until use. Theamount of FAAH in the crude extract was estimated to be 5% bySDS-PAGE.

For measuring FAAH inhibition, rat and mouse microsomeswere prepared from frozen brain tissue. Tissue was homogenized(3 � 15 s) in 20 mM phosphate buffer (pH = 7.4) with 5 mM EDTAand centrifuged (9,000g, 20 min). The soluble fraction (S9) wascollected and centrifuged again (100,000g, 1 h). The pelletcontaining microsomes was resuspended in 10 mM phosphatebuffer with 2.5 mM EDTA and 20% glycerol and kept frozen at�80 �C until use. Treatment with 50 nM of the potent FAAHinhibitor URB597 reduced OMP hydrolysis to less than 20%,indicating most if not all of the activity measured is from FAAH.

2.3. Fluorimetric enzyme assays

FAAH activity was measured in 0.1 M sodium phosphate bufferat pH = 8 with 0.1-0.2 mg/mL of bovine serum albumin (BSA) usingOMP as the substrate (Huang et al., 2007). Formation of thefluorescent methoxypyridine product was measured kinetically atlexcitation = 303 nm and lemission = 394 nm while reaction solutionswere kept at 37 �C. To determine IC50 values, final reactionsolutions contained approximately 0.9 mg of crude FAAH extract,[S]final = 50 mM and inhibitor in no more than 2% DMSO solution.Esterase activity (hCE1, hCE2 and AADAC) was measured under thesame conditions using CMNA as the substrate and measuring theliberation of 6-methoxynaphthaldehyde at lexcitation = 330 nm andlemission = 465 nm (Morisseau et al., 2009). All experiments wererun in either duplicate or triplicate and values reported asaverage � SD represent at least 3 independent experiments.

2.4. Cell culture, induction of adipocyte differentiation and chemicaltreatments

Murine 3T3-L1 fibroblasts (ATCC, Manassas, VA) were grownin Dulbecco's modified Eagle’s medium (DMEM) containing 10%calf serum (Hyclone) in a 5% CO2, 37 �C environment until theyreach confluence. To study the potentiating effects of parabensand other chemicals on adipocyte differentiation, the standarddifferentiation protocol was modified using a weaker GR agonistcortisone as previously described (Hu et al., 2013). Briefly, on theday of reaching confluence (designated as day 0), cells weretreated with DMEM containing 10% fetal bovine serum (FBS, AtlasBiologicals), 0.5 mM methylisobutylxanthine (MIX), 170 nM insu-lin and 5 mM cortisone for 3 days. The cells were then grown inmaintenance DMEM containing 10% FBS and 170 nM insulin foradditional two days until day 5 followed by in DMEM containing10% FBS until day 7.

Parabens and other chemicals were added in the differentiationmedia of 3T3-L1. Parabens and other chemicals were applied at

94 S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99

each change of the media at day 0, day 3, day 5 during the 7-dayprocess.

2.5. Oil O red staining and quantification

To quantify lipid accumulation, differentiated cells were fixedwith 4% paraformaldehyde overnight, then rinsed with deionizedwater and stained with Oil Red O solution (60% Oil Red O inisopropanol) for 10 min. After staining, the plates were rinsedwith deionized water. The images of the stained cell morphologywere taken by an integrated digital camera linked to a Micro-master inverted digital microscope (Thermo Fisher Scientific). Toquantify the staining, the Oil Red O was eluted with 100%isopropanol for 10 min and OD absorbance at 500 nm wasmeasured with GloMax-Multi Detection System (Promega,Madison, WI).

2.6. RT-PCR

Total RNA was isolated using TRI reagent (Molecular ResearchCenter Inc., Cincinnati, OH) according to the manufacturer’sinstructions. Total RNA abundance was quantified using aNanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies,Wilmington, DE). Reverse transcription was carried out using HighCapacity cDNA Reverse Transcription kits (Thermo Fisher Scientif-ic, Waltham, MA) according to the manufacturer’s instructions.mRNA expression of adipocyte marker genes and loading control36B4 were measured quantitatively using Absolute Blue QPCRSYBR Green ROX mix (Thermo Fisher Scientific, Waltham, MA).Primer Sequences are available upon request. PCR were run in a 96-well format using an ABI 7300HT instrument (Thermo FisherScientific, Waltham, MA). Cycle conditions were 50 �C 2 min, 95 �C15 min, then 40 cycles of 95 �C for 15 s/60 �C for 1 min. Relative geneexpression was calculated using the 2(�DDCt) method, whichnormalizes against 36B4.

3. Results

3.1. Parabens inhibit FAAH

The ability of several relevant parabens to inhibit FAAH wastested on a recombinant preparation of the enzyme and comparedagainst the well-characterized FAAH inhibitor URB597. Inhibitionpotencies (IC50) are reported in Table 1. Consistent with previousfindings, URB597 is highly potent, with a low nanomolar IC50, andits potency increases significantly over time (an order ofmagnitude better between 5 and 60 min pre-incubation), as itforms a covalent complex with the enzyme (Kathuria et al., 2003).Benzylparaben is one and two orders of magnitude more potentthan butyl- and propylparaben, respectively, while methylparabenis completely inactive. Additionally, the hydrolysis product, 4-hydroxybenzoic acid, is also completely inactive. After a 5 minincubation, benzylparaben is only 3-fold less potent than URB597.

Table 1Inhibition of fatty acid amide hydrolase (FAAH) by parabens and 4-hydroxybenzo-ate.

pre-incubation (min) 5 60Compound IC50 (mM)

URB597 0.040 � 0.011 0.0030 � 0.00054-HO-benzoate >100 >100Methylparaben >100 >100Propylparaben 9.4 � 1.0 17.1 � 1.8Butylparaben 1.1 � 0.2 2.4 � 0.5Benzylparaben 0.14 � 0.03 0.28 � 0.03

Compared to the time-dependent inhibition by URB597 that occursby a covalent intermediate, 1 h pre-incubation did not decrease theIC50 of parabens, indicating inhibition does not occur through acovalent intermediate. A more extensive analysis of time-depen-dence demonstrates potency does not appreciably change over anhour-long period (Fig. 1A). To explore its mechanism of inhibition,kinetic constants for benzylparaben were measured (Fig. 1B). Boththe apparent Vmax and the apparent Km changes as inhibitorconcentration increases and therefore the relationship betweenenzyme, substrate and inhibitor was analyzed as a linear, mixed-type model (Segel, 1993). This model derived a Ki value of52 � 14 nM with an a value of 9.7 � 6.6, indicating at saturatinginhibitor concentrations, the relative in vivo Km of AEA will changeby up to one order of magnitude.

To determine which structural features confer potency,several analogs of benzylparaben were synthesized and tested(Table 2). Modification of the phenol to the methylaminophenyl(1) decreases potency 100-fold, while modification to the anisole(2), phenyl (3) or nitrophenyl (4) completely removed potencytowards FAAH. This general trend suggests the p-hydroxyl is anessential component for activity against FAAH. Replacement of thebenzyl alcohol with cyclohexyl-methanol (5) decreased potency2-fold while replacement to the heptylparaben (6) had nochange in potency, indicating the potency is primarily determinedby the hydrophobicity of the alcohol portion of the ester.Finally, modification of the ester to an amide (7) reduced potency100-fold, indicating the ester is an essential component forpotency.

Fig. 1. Enzyme kinetics associated with FAAH inhibition by butylparaben andbenzylparaben. (A) Inhibition of FAAH is independent on incubation time of enzymewith either butylparaben or benzylparaben. (B) Benzylparaben inhibits FAAHthrough a mixed type mechanism as evidenced by a change in Vmax

app (5.5 RFU/s to2.0 RFU/s) and KM

app (18 mM to 98 mM). The calculated Ki and a assuming a linearmixed-type model of inhibition is 52 � 14 nM and 9.7 � 6.6, respectively.

Table 2Structure activity relationships of paraben-liked molecules on fatty acid amidehydrolase (FAAH) inhibition.

Compound R1 R2 X IC50 (mM)(5 min pre-incubation)

Benzylparaben O 0.14 � 0.03

1 O 18.2 � 4.9

2 O >100

3 O >100

4 O >100

5 O 0.36 � 0.03

6 O 0.15 � 0.047 N 8.6 � 1.3

S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99 95

3.2. Inhibition by parabens is selective for FAAH and is not species-dependent

To test the possibility that parabens inhibit other serinehydrolases, the potencies of butylparaben and benzylparabenwere compared with recombinant preparations of human carbox-ylesterase 1 and 2 (hCE1 and hCE2) and arylacetamide deacetylase(AADAC) using the general esterase substrate CMNA (Shan andHammock, 2001) (Table 3). For all enzymes, the IC50 of bothparabens are above 10 mM (the highest concentration tested),suggesting that it is unlikely that parabens inhibit serine hydro-lases in general.

In addition to selectivity against other enzymes, potency ofparabens was compared between human and rodent FAAH. Forrodent data, brain microsome preparations were used instead of

Table 3Relative selectivity of butylparaben and benzylparaben on FAAH in rodent speciesand on other recombinant enzymes.

Enzyme Source IC50 (mM) (5 min pre-incubation)

URB597 Butylparaben Benzylparaben

Human Recombinant hCE1b 2.2 >10 >10Human Recombinant hCE2b 0.73 >10 >10Human Recombinant AADACb >10 >10 >10Human Recombinant FAAHa 0.040 � 0.011 1.1 � 0.2 0.14 � 0.03Mouse Brain Microsomesa <0.001 1.6 0.21Rat Brain Microsomesa 0.005 1.1 0.23

a Measured by hydrolysis of OMP, [S] = 50 mM.b Measured by hydrolysis of CMNA, [S] = 50 mM.

recombinant enzyme preparations. URB597 was used as a controlto confirm the same fluorescence-based substrate could be usedwith brain microsome preparations. URB597 inhibits over 80% ofthe activity at 50 nM in both rat and mouse microsomes,demonstrating FAAH accounts for most of N-(6-methoxypyridin-3-yl) octanamide (OMP) hydrolysis and the IC50 from thesepreparations represents approximate values of parabens on mouseand rat FAAH. The values for both benzylparaben and butylparabenhad a less than 2-fold difference between human and both rodentspecies (Table 3).

3.3. The effects of FAAH inhibition by parabens and other compounds,and interaction with AEA on 3T3-L1 adipocyte differentiation

To determine whether previously observed actions of parabenson pre-adipocytes are mediated by FAAH inhibition (Hu et al.,2013), we first tested whether FAAH inhibition could plausiblyinduce differentiation. Although two inhibitors, URB597 andPF622, were tested on pre-adipocytes, only exposure to 10 mMof URB597 resulted in increased lipid accumulation as shown by OilRed O staining (Fig. 2A). This stimulation was supported byincreased PPARg, C/EBPa and FABP4 mRNA expression (Fig. 2B),which are common biomarkers of adipocyte differentiation. Aspreviously observed, butylparaben and benzylparaben dose-dependently enhanced 3T3-L1 adipocyte differentiation, as shownby increased Oil Red O staining and enhanced PPARg, C/EBPa andFABP4 mRNA expression.

If these effects were mediated through FAAH inhibition, wespeculated the biologically active substrates of FAAH wouldenhance paraben- or other FAAH inhibitor-induced differentiation.Paraben, PF622, or URB597 did not further increase AEA-mediatedOil Red O staining (Fig. 3A). Neither co-treatment of AEA withparaben, PF622, or URB597 further enhanced adipocyte markergene expression (Fig. 3B). This could indicate that a FAAH substratedifferent from AEA might account for the effects of the FAAHinhibitors or the lack of additive effect could be due to saturation ofAEA in the cell system.

3.4. 3T3-L1 adipocyte differentiation by FAAH inhibitors is CB1R-independent

Most of the physiologic effects of FAAH inhibitors are believedto be mediated through activation of one or both of thecannabinoid receptors (CB1 and CB2). To test whether the effectsof parabens on adipocyte differentiation were mediated in total orin part by the CB1 receptor, we performed a co-treatment ofbutylparaben, benzylparabens and AEA with the potent CB1

antagonist rimonabant (Rinald-Carmona et al., 1994) (Fig. 4).None of the tested concentrations of rimonabant reduced Oil Red Ostaining in either the AEA or paraben groups, although highestdose of rimonabant reduced C/EBPa expression in the butylpar-aben group. Interestingly, rimonabant treatment seemed toactually result in a robust increase in several markers, includingC/EBPa expression in the AEA group, PPARg and C/EBPaexpression in the benzylparaben group and FABP4 in thebutylparaben group (Fig. 4B).

4. Discussion

In this study, we identified benzylparaben as a mixed-type,time-independent inhibitor of fatty acid amide hydrolase. Otherparabens, including butylparaben, also inhibit FAAH, althoughtheir potency decreases with decreasing chain length. Wecalculated the Ki of benzylparaben towards FAAH to be52 � 14 nM and the a value, which indicates the maximum fold-difference possible to the Km or Ki, was 9.7 �6.6. Since substrates

Fig. 2. 3T3-L1 cells were induced to differentiate in the presence of PF622, URB597, butylparaben (BuP), or benzylparaben (BnP) (1, 10, 50 mM) for 7 days. (A) Parabens dosedependently increase adipocyte differentiation as measured by oil red O while URB597 increases differentiation at 10 mM. (B) Parabens and URB597 increase mRNAexpression of the adipocyte markers PPARg, C/EBPa and FABP4 (n = 3).

96 S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99

for FAAH, such as AEA, are produced on demand in response tostimuli (Piomelli and Sasso, 2014), this means at a basal state therelative binding affinity of benzylparabens for FAAH will berelatively high but as substrates are produced, the relative affinitywill decrease. However, since the interaction is mixed-type insteadof competitive, the relative Ki will never be higher than 1 mM.Isoflavones, including Biochanin A, have been reported to inhibitFAAH in a similar mixed-type fashion (Thors et al., 2010). Several ofthese isoflavones have similar structural features including a paraphenol to the carbonyl group. Additionally, both parabens andbiochanin A have potencies that are comparable between humanand rodent species. Given these similarities, it is possible parabensand biochanin A may interact with a well-conserved binding siteon FAAH close to but distinct from the active site that could be usedin future endeavors for designing novel inhibitors.

Since parabens were previously reported to enhance adipocytedifferentiation with a comparable structure-activity relationship

(Hu et al., 2013), we hypothesized that differentiation could bemediated by FAAH inhibition. Here, we tested adipogenic effects oftwo FAAH inhibitors, PF622 and URB597 alongside with parabensat dose range of 1–50 mM. We found only URB597 increaseddifferentiation and only at a concentration of 10 mM, not at 50 mM,and a weaker FAAH inhibitor PF622 had no effects at any of thedoses tested. Both of these are in contrast to the dose-dependentadipogenic effects of parabens. The endogenous FAAH substratearachidonoyl ethanolamide (AEA) was reported to stimulateadipocyte differentiation of 3T3-L1 cells (Bouaboula et al., 2005)and rat primary preadipocytes (Karaliota et al., 2009). In ratpreadipocytes, adipogenic effects of AEA were inhibited by FAAHinhibitor URB597 (3 mM) and the COX-2 inhibitor indomethacin,suggesting that adipogenic effects of AEA might be due to the AEAmetabolites derived from both FAAH and COX-2 (Karaliota et al.,2009). Here, we found AEA increased expression of severalmarkers of differentiation (PPARg and C/EBPa) but its effects

Fig. 3. 3T3-L1 cells were pretreated with 10 mM of PF622, URB597, or parabens for two hours. The cells were then induced to differentiate in the presence ofarachidonoylethanolamide (AEA) (10 mM) or vehicle (absolute alcohol, EtOH) while maintained in 10 mM of PF622, URB597, butylparaben (BuP), benzylparaben (BnP) orvehicle (DMSO) for 7 days. (A) AEA and parabens, PF622 or URB597 do not synergize to increase adipocyte differentiation as measured by oil red O. (B) AEA does not affect theregulation of mRNA expression of the adipocyte markers PPARg, C/EBPa and FABP4 in the presence or absence of parabens or URB597 (n = 3). *p < 0.05 compared to therespective control within the treatment.

S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99 97

were not significantly changed with FAAH inhibition by URB597,butylparaben or benzylparaben in 3T3-L1 cells. Although nosignificant changes were observed from FAAH inhibition whenURB597 or paraben were added to AEA-treated cells, this may bedue to saturation of AEA in the cell system. On the other hand, AEA-induced 3T3-L1 adipocyte differentiation was reported to bedependent on direct binding and activation of PPARg (Bouaboulaet al., 2005). Therefore, it remains to be determined whether AEApromotes adipocyte differentiation via its metabolites from FAAH(or COX-2) or itself as a PPARg agonist. To test further whetherdifferentiation could be due to CB1R activation, we treated 3T3-L1cells with rimonabant, a CB1R antagonist, and found both AEA andparabens’ adipogenic effects are independent of CB1R activation.Activation of the CB1 receptor stimulates adipogenesis (Bellocchioet al., 2008) while CB2 activation attenuates adipogenesis (Vertyet al., 2015; Rossi et al., 2016); thus, the CB2 receptor is unlikely to

be responsible for the paraben-enhanced adipogenesis. However,the fact that CB2 receptor was not excluded as a potential target is alimitation of our study and this hypothesis should be examined infuture studies. Taken together, our results suggest that adipogeniceffects of FAAH inhibition by parabens or other inhibitors in 3T3-L1cells may be due to accumulation of AEA, leading to more PPARgactivation.

Although studies have implicated parabens in endocrinedisruption (Chen et al., 2007), their use in cosmetics has beenconsidered safe by the United States (U.S. FDA, 2007). This is due, inpart, to their low metabolic stability and fast excretion with 81–85% excreted in the urine after the first 24 h and over half of thatexcreted as p-hydroxyhippuric acid, the primary metabolite (Mooset al., 2015). Despite the rapid metabolism, the high prevalence ofthese products may result in regular daily exposure, as evidencedby a high incidence of detection in urine (Ye et al., 2006; Tefre de

Fig. 4. 3T3-L1 cells were pretreated with Rimonabant (0.1, 0.2, or 0.5 mM) or vehicle (DMSO) for two hours. Then, the cells were induced to differentiate in the presence of AEA(10 mM), butylparaben (50 mM) or benzylparaben (50 mM), while maintained in rimonabant for 7 days. (A) Increasing concentrations of rimonabant had no effect on paraben-enhanced adipocyte differentiation as measured by oil red O. (B) Effects of rimonabant on parabens-mediated upregulation of adipocyte marker PPARg, C/EBPa, and FABP4.The dark triangles indicate the increasing doses of rimonabant shown in (A). *p < 0.05 compared with the control (0 mM of rimonabant) within each treatment group. NS, notsignificant.

98 S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99

Renzy-Martin et al., 2014). In a survey of personal care products,methylparaben is used at the highest concentrations, while propyl-and butylparaben are regularly used but at lower concentrationsand benzylparaben is rarely used (Guo and Kannan, 2013). Thisgeneral trend approximately matches the relative concentrationsof individual parabens in the urine (Ye et al., 2006; Tefre de Renzy-Martin et al., 2014). Urinary concentrations of butylparaben reach0.1–0.5 mM (Calafat et al., 2010; Tefre de Renzy-Martin et al., 2014),which is in the range relevant to FAAH inhibition but not adipocytedifferentiation. While benzylparaben is more biologically active onFAAH, butylparaben is more relevant from the perspective ofhuman exposure.

It is difficult to determine, based on human exposure, whetherthe potencies of these parabens toward FAAH and adipocytedifferentiation are substantial enough to cause major physiologicchanges. Although the potent FAAH inhibitor PF-04457845 waswell tolerated in the clinic with no major adverse side effects(Huggins et al., 2012), the death of a volunteer in a Phase I trial ofthe FAAH inhibitor BIA 10-2474 has increased safety concernsrelated to FAAH inhibition (von Schaper, 2016). While the potencyof parabens is relatively weak compared to drugs optimized totarget FAAH, the efficacy of FAAH inhibition may be enhanced byother factors. Recent reports have demonstrated FAAH inhibitionsynergizes with other targets including soluble epoxide hydrolase(sEH) (Sasso et al., 2015) and cyclooxygenase (COX) (Naidu et al.,

2009). While these studies investigate synergy in the context ofnociception, both sEH and COX are generally relevant to adipo-genesis and the regulation of inflammation (Karaliota et al., 2009;De Taeye et al., 2010). Triclocarban, a common antibacterial foundin consumer care products, similarly inhibits sEH (Schebb et al.,2011). Thus, while exposure to either triclocarban or parabensalone may not be substantial enough to have a major impact onhuman health, combined exposure to triclocarban and eitherbutyl- or benzylparaben could have significant biological effects.Similarly, cyclooxygenase inhibitors, including aspirin, remainhighly used pharmaceuticals whose potency may be enhanced dueto synergism with parabens. Further studies will be necessary tovalidate whether synergism occurs from exposure to thesecommonly used consumer and pharmaceutical products.

5. Conclusion

In conclusion, parabens, preservatives found in personal careproducts, inhibit the endocannabinoid enzyme FAAH. Based on thestructure activity relationship, activity requires a para-hydroxybenzoic ester and potency increases as hydrophobicity increases.While FAAH inhibition may account for paraben-enhancedadipocyte differentiation, the adipogenic activity is independentof the CB1 receptor. Further experiments are needed to determinewhether FAAH inhibitor activities are required for parabens’

S.D. Kodani et al. / Toxicology Letters 262 (2016) 92–99 99

adipogenic effects. Moreover, further experiments are needed todetermine whether doses relevant to human exposure can elicitphysiologic effects in vivo.

Acknowledgements

This work was supported by the National Institute ofEnvironmental Health Sciences grant R01 ES002710 and theNational Institute of Environmental Health Sciences SuperfundResearch Program Grant P42 ES004699. SDK was supported by aNIGMS-funded Pharmacology Training Program (T32GM099608).

Notes: SDK and BDH are authors on a patent disclosure filedthrough the University of California, Davis for fatty acid amidehydrolase inhibitors that are unrelated to the work described here.

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.toxlet.2016.09.011.

References

Bellocchio, L., Cervino, C., Vicennati, V., Pasquali, R., Pagotto, U., 2008. Cannabinoidtype 1 receptor: another arrow in the adipocytes’ bow. J. Neuroendocrinol. 20(Suppl. 1), 130–138.

Bledzka, D., Gromadzinska, J., Wasowicz, W., 2014. Parabens. From environmentalstudies to human health. Environ. Int. 67, 27–42.

Bouaboula, M., Hilairet, S., Marchand, J., Fajas, L., Le Fur, G., Casellas, P., 2005.Anandamide induced PPARgamma transcriptional activation and 3T3-L1preadipocyte differentiation. Eur. J. Pharmacol. 517, 174–181.

Bowles, N.P., Karatsoreos, I.N., Li, X., Vemuri, V.K., Wood, J.A., Li, Z., Tamashiro, K.L.,Schwartz, G.J., Makriyannis, A.M., Kunos, G., Hillard, C.J., McEwen, B.S., Hill, M.N.,2015. A peripheral endocannabinoid mechanism contributes to glucocorticoid-mediated metabolic syndrome. Proc. Natl. Acad. Sci. U. S. A. 112, 285–290.

Calafat, A.M., Ye, X., Wong, L.Y., Bishop, A.M., Needham, L.L., 2010. Urinaryconcentrations of four parabens in the U.S. population: NHANES 2005–2006.Environ. Health Perspect. 118, 679–685.

Chen, J., Ahn, K.C., Gee, N.A., Gee, S.J., Hammock, B.D., Lasley, B.L., 2007.Antiandrogenic properties of parabens and other phenolic containing smallmolecules in personal care products. Toxicol. Appl. Pharmacol. 221, 278–284.

Christopoulou, F.D., Kiortsis, D.N., 2011. An overview of the metabolic effects ofrimonabant in randomized controlled trials: potential for other cannabinoid 1receptor blockers in obesity. J. Clin. Pharm. Ther. 36, 10–18.

De Taeye, B.M., Morisseau, C., Coyle, J., Covington, J.W., Luria, A., Yang, J., Murphy, S.B., Friedman, D.B., Hammock, B.B., Vaughan, D.E., 2010. Expression andregulation of soluble epoxide hydrolase in adipose tissue. Obesity (Silver,Spring) 18, 489–498.

FDA, 1973. Methyl paraben and propyl paraben. Affirmation of GRAS status of directhuman food ingredients. Fed. Regist. 38, 20048–20050.

Guo, Y., Kannan, K., 2013. A survey of phthalates and parabens in personal careproducts from the United States and its implications for human exposure.Environ. Sci. Technol. 47, 14442–14449.

Hoberman, A.M., Schreur, D.K., Leazer, T., Daston, G.P., Carthew, P., Re, T., Loretz, L.,Mann, P., 2008. Lack of effect of butylparaben and methylparaben on thereproductive system in male rats. Birth Defects Res B Dev. Reprod. Toxicol. 83,123–133.

Hu, P., Chen, X., Whitener, R.J., Boder, E.T., Jones, J.O., Porollo, A., Chen, J., Zhao, L.,2013. Effects of parabens on adipocyte differentiation. Toxicol. Sci. 131, 56–70.

Huang, H., Nishi, K., Tsai, H.J., Hammock, B.D., 2007. Development of highly sensitivefluorescent assays for fatty acid amide hydrolase. Anal. Biochem. 363, 12–21.

Huggins, J.P., Smart, T.S., Langman, S., Taylor, L., Young, T., 2012. An efficientrandomised, placebo-controlled clinical trial with the irreversible fatty acidamide hydrolase-1 inhibitor PF-04457845, which modulates endocannabinoidsbut fails to induce effective analgesia in patients with pain due to osteoarthritisof the knee. Pain 153, 1837–1846.

Karaliota, S., Siafaka-Kapadai, A., Gontinou, C., Psarra, K., Mavri-Vavayanni, M., 2009.Anandamide increases the differentiation of rat adipocytes and causesPPARgamma and CB1 receptor upregulation. Obesity (Silver Spring) 17, 1830–1838.

Kathuria, S., Gaetani, S., Fegley, D., Valino, F., Duranti, A., Tontini, A., Mor, M., Tarzia,G., La Rana, G., Calignano, A., Giustino, A., Tattoli, M., Palmery, M., Cuomo, V.,Piomelli, D., 2003. Modulation of anxiety through blockade of anandamidehydrolysis. Nat. Med. 9, 76–81.

Moos, R.K., Angerer, J., Dierkes, G., Bruning, T., Koch, H.M., 2015. Metabolism andelimination of methyl, iso- and n-butyl paraben in human urine after single oraldosage. Arch. Toxicol. doi:http://dx.doi.org/10.1007/s00204-015-1636-0.

Morisseau, C., Merzlikin, O., Lin, A., He, G., Feng, W., Padilla, I., Denison, M.S., Pessah,I.N., Hammock, B.D., 2009. Toxicology in the fast lane: application of high-throughput bioassays to detect modulation of key enzymes and receptors.Environ. Health Perspect. 117, 1867–1872.

Naidu, P.S., Booker, L., Cravatt, B.F., Lichtman, A.H., 2009. Synergy between enzymeinhibitors of fatty acid amide hydrolase and cyclooxygenase in visceralnociception. J. Pharmacol. Exp. Ther. 329, 48–56.

Nishi, K., Huang, H., Kamita, S.G., Kim, I.H., Morisseau, C., Hammock, B.D., 2006.Characterization of pyrethroid hydrolysis by the human liver carboxylesteraseshCE-1 and hCE-2. Arch. Biochem. Biophys. 445, 115–123.

Pereira-Fernandes, A., Demaegdt, H., Vandermeiren, K., Hectors, T.L., Jorens, P.G.,Blust, R., Vanparys, C., 2013. Evaluation of a screening system for obesogeniccompounds: screening of endocrine disrupting compounds and evaluation ofthe PPAR dependency of the effect. PLoS One 8, e77481.

Pertwee, R.G., 2015. Endocannabinoids and their pharmacological actions. Handb.Exp. Pharmacol. 231, 1–37.

Piomelli, D., Sasso, O., 2014. Peripheral gating of pain signals by endogenous lipidmediators. Nat. Neurosci. 17, 164–174.

Rinald-Carmona, M., Barth, F., Héaulme, M., Shire, D., Calandra, B., Congy, C.,Martinez, S., Maruani, J., Néliat, G., Caput, D., Ferrara, P., Soubrié, P., Brelière, J.C.,Fur, G.L., 1994. SR141716A, a potent and selective antagonist of the braincannabinoid receptor. FEBS Lett. 350, 240–244.

Rossi, F., Bellini, G., Luongo, L., Manzo, I., Tolone, S., Tortora, C., Bernardo, M.E.,Grandone, A., Conforti, A., Docimo, L., Nobili, B., Perrone, L., Locatelli, F., Maione,S., Del Giudice, E.M., 2016. Cannabinoid receptor 2 as anti-obesity target:inflammation, fat storage and browning modulation. J. Clin. Endocrinol. Metab.(jc20154381).

Sasso, O., Wagner, K., Morisseau, C., Inceoglu, B., Hammock, B.D., Piomelli, D., 2015.Peripheral FAAH and soluble epoxide hydrolase inhibitors are synergisticallyantinociceptive. Pharmacol. Res. 97, 7–15.

Schebb, N.H., Inceoglu, B., Ahn, K.C., Morisseau, C., Gee, S.J., Hammock, B.D., 2011.Investigation of human exposure to triclocarban after showering andpreliminary evaluation of its biological effects. Environ. Sci. Technol. 45, 3109–3115.

Segel, I.H., 1993. Enzyme Kinetics: Behavior and Analysis of Rapid Equilibrium andSteady-State Enzyme Systems. Wiley, New York.

Shan, G., Hammock, B.D., 2001. Development of sensitive esterase assays based onalpha-cyano-containing esters. Anal. Biochem. 299, 54–62.

Silvestri, C., Di Marzo, V., 2013. The endocannabinoid system in energy homeostasisand the etiopathology of metabolic disorders. Cell Metab 17, 475–490.

Soni, M.G., Carabin, I.G., Burdock, G.A., 2005. Safety assessment of esters of p-hydroxybenzoic acid (parabens). Food Chem. Toxicol. 43, 985–1015.

Taxvig, C., Dreisig, K., Boberg, J., Nellemann, C., Schelde, A.B., Pedersen, D.,Boergesen, M., Mandrup, S., Vinggaard, A.M., 2012. Differential effects ofenvironmental chemicals and food contaminants on adipogenesis, biomarkerrelease and PPARgamma activation. Mol. Cell. Endocrinol. 361, 106–115.

Tefre de Renzy-Martin, K., Frederiksen, H., Christensen, J.S., Boye Kyhl, H.,Andersson, A.M., Husby, S., Barington, T., Main, K.M., Jensen, T.K., 2014. Currentexposure of 200 pregnant Danish women to phthalates, parabens and phenols.Reproduction 147, 443–453.

Thors, L., Burston, J.J., Alter, B.J., McKinney, M.K., Cravatt, B.F., Ross, R.A., Pertwee, R.G., Gereau, R.W., Wiley, J.L., Fowler, C.J., 2010. Biochanin A, a naturally occurringinhibitor of fatty acid amide hydrolase. Br. J. Pharmacol. 160, 549–560.

Verty, A.N., Stefanidis, A., McAinch, A.J., Hryciw, D.H., Oldfield, B., 2015. Anti-obesityeffect of the CB2 receptor agonist JWH-015 in diet-induced obese mice. PLoSOne 10, e0140592.

Vida, M., Rivera, P., Gavito, A.L., Suarez, J., Pavon, F.J., Arrabal, S., Romero-Cuevas, M.,Bautista, D., Martinez, A., de Fonseca, F.R., Serrano, A., Baixeras, E., 2014. CB1blockade potentiates down-regulation of lipogenic gene expression in perirenaladipose tissue in high carbohydrate diet-induced obesity. PLoS One 9, e90016.

Wang, Q.A., Scherer, P.E., Gupta, R.K., 2014. Improved methodologies for the study ofadipose biology: insights gained and opportunities ahead. J. Lipid Res. 55, 605–624.

Ye, X., Bishop, A.M., Reidy, J.A., Needham, L.L., Calafat, A.M., 2006. Parabens asurinary biomarkers of exposure in humans. Environ. Health Perspect.114, 1843–1846.

von Schaper, E., 2016. Bial incident raises FAAH suspicions. Nat. Biotechnol. 34, 223.

Supplementary Materials:

General synthetic procedure for 4-substituted benzyl benzoates (1-4). Benzyl alcohol (1

mmol) was added to 1 equivalent each of 4-substituted benzoic acid, 1-ethyl-3-(3-

dimethylaminopropyl)carbodiimide (EDCI) and 4-dimethylaminopyridine (DMAP) in 5 mL of

dichloromethane. The reaction was stirred overnight at room temperature and stopped by

addition of 0.2 M HCl. Product was extracted from the aqueous layer by two washes with ethyl

acetate (EtOAc) and dried over MgSO4. Solvent was evaporated in vacuo and purified by flash

chromatography with 7:3 Hexane:EtOAc.

Synthesis of benzyl 4-(methylamino)benzoate (1). Final product was a yellowish oil. 1H NMR

(300 MHz, DMSO- d6) δ 2.70 (d, J = 4.8 Hz, 3H), 5.21-5.31 (s, 2H), 6.50-6.62 (m, 3H), 7.27-

7.47 (m, 5H), 7.73 (d, J = 7.5 Hz, 2H).

Synthesis of benzyl 4-methoxybenzoate (2). Final product was a yellowish oil. 1H NMR (300

MHz, DMSO- d6) δ 3.79 (s, 3H), 5.29 (s, 2H), 7.01 (d, J = 9.0 Hz, 2H), 7.25-7.49 (m, 5H), 7.93

(d, J = 8.7 Hz, 2H).

Synthesis of benzyl benzoate (3). Final product was a yellowish oil. 1H NMR (300 MHz,

DMSO-d6) δ 8.01 (d, J = 6.3 Hz, 2H), 7.67 (t, J = 7.4 Hz, 1H), 7.58 – 7.29 (m, 6H), 5.36 (s, 2H).

Synthesis of benzyl 4-nitrobenzoate (4). Final product was a yellow solid that was

recrystallized in ethyl acetate and hexane. 1H NMR (300 MHz, DMSO- d6) δ 5.40 (s, 2H), 7.34-

7.55 (m, 2H) 8.22 (dt, J = 9 Hz, J = 2.7 Hz, 2H), 8.35 (dt, J = 9 Hz, J = 2.7 Hz, 2H). MP = 84.3-

85.4 °C (84.7 °C).

General synthetic procedure for 4-hydroxy 1-substituted benzoates (5-6). 4-hydroxybenzoate

was protected by addition of imidazole, and tert-butyl diphenyl silyl chloride in

dimethylformamide (DMF). The reaction was stirred overnight. Appropriate alcohol (1 mmol)

was added to 1 equivalent each of 4-substituted benzoic acid, EDCI and DMAP in 5 mL of

dichloromethane. The reaction was stirred overnight at room temperature and stopped by

addition of 0.2 M HCl. Product was extracted from the aqueous layer by two washes with EtOAc

and dried over MgSO4. Solvent was evaporated in vacuo and purified by flash chromatography

with 7:3 Hexane:EtOAc.

Products were then deprotected in tetrahydrofuran (THF) (2 mL) containing 1.1 equivalent of

TBAF. The resulting product was diluted in water, extracted with EtOAc, and dried with MgSO4.

The final product was purified by flash chromatography.

Synthesis of cyclohexylmethyl 4-hydroxybenzoate (5). Final product was a white solid that

was recrystallized in water and methanol. 1H NMR (300 MHz, DMSO- d6) δ 0.92-1.35 (m, 5H),

1.57-1.85 (m, 6H), 3.96-4.09 (d, J = 6 Hz, 2H), 6.85 (d, J = 8.7 Hz, 2H), 7.81 (d, J = 8.6, 2H),

10.31 (s, 1H). MP = 135.3-136.1 °C (135.6 °C).

Synthesis of N-benzyl-4-hydroxybenzamide (7). Final product was a white solid that was

recrystallized in hexane and ethyl acetate. 1H NMR (300 MHz, DMSO- d6) δ 4.43 (d, J = 6.0 Hz,

2H), 6.79 (d, J = 8.7 Hz, 2H), 7.14-7.37 (m, 5H), 7.75 (d, J = 8.7 Hz, 2H), 8.76 (t, J = 6 Hz, 1H),

9.95 (s, 1H). MP = 167.6-168.6 °C (135.6 °C).