Cc

ESa

b

c

a

ARRAA

KCRSGNE

1

ttamcoamasS

h0

Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

j o ur na l ho mepage: www.elsev ier .com/ locate / jpba

hemical analysis and potential endocrine activities of aluminiumoatings intended to be in contact with cosmetic water

lias Bou-Marouna,∗, Laurence Dahbib, Marie-Pierre Gomez-Berradac, Philippine Pierrec,andrine Rakotomalalac, Pierre-Jacques Ferretc, Marie-Christine Chagnonb

Univ. Bourgogne Franche-Comté, AgroSup Dijon, PAM UMR A 02.102, Procédés Alimentaires et Microbiologiques, 1 Esplanade Erasme, 21000 Dijon, FranceDerttech “Packtox”, NUTOX, INSERM U1231, Univ. Bourgogne Franche-Comté, AgroSup Dijon, 1 Esplanade Erasme, 21000 Dijon, FranceCentre de R&D Pierre Fabre Dermo-Cosmétique, Département Sécurité Produits, 3, avenue Hubert Curien, BP 13562, 31035 Toulouse, France

r t i c l e i n f o

rticle history:eceived 8 April 2017eceived in revised form 23 July 2017ccepted 31 July 2017vailable online 1 August 2017

eywords:osmetic aluminium coatingeporter gene bioassayolid phase extractionC/HPLC–MSon-intended added substancesndocrine activities

a b s t r a c t

The objective of the work was to check the presence of Non-Intended Added Substances (NIAS) withhormonal activities in aluminium coatings extracts coded: AA, BBF, MC and RR, furnished by four differ-ent suppliers. Water samples were prepared at room temperature or at 40 ◦C for three months to verifythe storage effect on the coatings. Solid phase extraction was used to concentrate and to extract coat-ing substances. Hormonal activities were checked in vitro using reporter gene bioassays. Except BBF, allextracts induced a weak but significant estrogenic agonist activity in the human cell line. Using an estro-genic antagonist (ICI-182, 780), the answer was demonstrated specific in the bioassay. RR was the onlyextract to induce a concentration dependent anti-androgenic response in the MDA-KB2 cell line. Analy-sis performed using GC–MS and HPLC–MS detected 12 substances in most of the extracts. 8 NIAS werepresent. Among them, 4 were identified with certainty: HMBT, BGA, DCU and BPA. Estrogenic potencywas BPA > DCU > BGA > HMBT. HMBT was also anti-androgenic at high concentration. Combining chem-ical analysis and bioassays data, we demonstrated that in the RR and the RR40 extracts, the observedestrogenic response was mainly due to BPA, the anti-androgenic activity of RR could be due to a syner-

gism between HMBT and BPA. For MC and AA, estrogenic responses appear to be due to the presenceof DCU. Except BBF, storage conditions tended to increase estrogenic activities in all extracts. However,in term of risk assessment, activities observed were negligible. This work demonstrated that sensitivebioassays are pertinent tools in complement to chemical analysis to monitor and check the presence of

ity in

NIAS with hormonal activ

. Introduction

Materials used for the packaging play a beneficial role to pro-ect cosmetic products against moisture, light, microorganism ofhe different formulations used in cosmetology. Materials qualitynd safety are also major issues of concern for packaging and cos-etic companies, for risk assessors, for regulatory agencies and for

onsumers. Regards to Food Contact Materials (FCM) dependingn physical/chemical parameters and chemical composition, pack-ging may transfer constituents to foods (a phenomenon calledigration) resulting in small, but measurable human exposures. In

ddition to substances of known origin (Intentionally Added Sub-tance or IAS), materials may also release Non-Intentionally Addedubstances (NIAS) which represent a great part of all migrating

∗ Corresponding author.E-mail address: elias.bou-maroun@agrosupdijon.fr (E. Bou-Maroun).

ttp://dx.doi.org/10.1016/j.jpba.2017.07.061731-7085/© 2017 Elsevier B.V. All rights reserved.

coating extracts.© 2017 Elsevier B.V. All rights reserved.

substances [1,2]. The most common description of NIAS is «forestof peaks» appearing in chromatograms, comprising: oligomersand their by-products, minor impurities, degradation products ofIAS during packaging processing and/or packaging storage [2].Due to the interaction of the packaging and the matrix, riskmanagement on NIAS is a major challenge for polymer indus-tries, packaging converters and cosmetic industries. Developingscientifically-based guidance and rules is a key priority for theindustry and for protecting consumers. However, mitigating therisks for NIAS (identification, minimization or avoidance) is com-plicated by the absence of rules and clear identification of themain sources (impurities, cross-contamination, residues). Concern-ing analytical methods for NIAS identification: I) The formulationof materials is constantly evolving, adapting to novel uses, givingrise to changes in its composition with a large number of substances

(IAS and NIAS) being present in the extract, rendering an exhaustiveidentification virtually unattainable [3]. II) The analytical difficultyresides in very low concentrations of NIAS (ppb level) which chal-

6 tical an

latTuwCAatcpibf

odstfacmeoojctt“e[

accscbganm

dcmstrgiuH

2

2

f4B

42 E. Bou-Maroun et al. / Journal of Pharmaceu

enges analytical detection limits. Furthermore, reaction schemesre complex and depend on specific process conditions makingheir prediction in undefined chemical mixtures not exhaustive.he (EC) regulation No 1223/2009 specifies that «a cosmetic prod-ct made available on the market shall be safe for human healthhen used under normal or reasonably foreseeable conditions».ommission Implementing Decision 2013/674/EC on guidelines onnnex I to (EC) Regulation (EU) No 1223/2009 gives further guid-nce regards to the information about the packaging material andhe potential release of substances from the packaging. Indeed, theombination of packaging material, formulation of the cosmeticroduct and contact with the external environment may have an

mpact on the safety of the finished product, due to interactionetween the product and the packaging material, substance releaserom the packaging material.

In term of toxicological hazard, Endocrine Disruption (ED) isf high concern as an emerging public health risk in differentomains, as demonstrated with BPA in FCM. There is also growingcientific, media and public attention regarding packaging migra-ion as a source of chemical contaminations: chemicals leachingrom packaging (e.g.: bisphenols or phthalates) in matrixes suchs food contributing to human EDs exposure [4]. Indeed, chemi-als exhibiting Estrogenic Activity (EA) for example can produceany biological and adverse health effects in mammals such as

arly puberty in females, reduced sperm counts, altered functionsf reproductive organs, obesity and increased rates of some breast,varian, testicular, and prostate cancers [5]. Foetal, new-born, anduvenile mammals are reported to be particularly sensitive tohemicals having EA, effects being observed at very low doses. Fur-hermore, toxicological interactions between chemicals in relationo dose additivity and synergism or supra additivity (also calledcocktail effects”) are also of major concern. These effects at lowxposure cannot be excluded notably with endocrine disruptors6].

Then, dealing with NIAS and risk assessment, it is increasinglycknowledged that a traditional approach based on the identifi-ation/quantification of all substances with their full toxicologicalharacterization is impractical, highly resource intensive, time con-uming and undesirable as many NIAS are likely to be of no safetyoncern [7]. One way to get around these drawbacks is to useioassays in complement to analytical methods. With FCM, strate-ies have already been set up combining analytical methodologiesnd bioassays, demonstrating the relevance of bioassays as perti-ent tools to address the risk associated to unpredicted unknownigrants [8].The objective of this work was to check the presence of unpre-

icted NIAS with hormonal activities in extracts of aluminiumoatings used for water cosmetic products performing combinedethodologies (chemical analysis and bioassays) in order to insure

afety and quality of the cosmetic products intended to be in con-act with these aluminium coatings. The storage time effect on theelease of NIAS was also monitored. Reliable and sensitive reporterene bioassays using human cell lines were performed on extractsn parallel to chemical analysis using gas chromatography and liq-id chromatography coupled to mass spectrometry, GC–MS andPLC–MS.

. Material and methods

.1. Chemicals and instrumentation

All organic solvents were HPLC grade and were purchasedrom Sigma Aldrich, France. The chromatographic standards,,4′-(2,2-propandiyl)diphenol (BPA, CAS n◦ 80-05-7), 2-(1,3-enzothiazol-2-ylsulfanyl)ethanol (HMBT, CAS n◦ 4665-63-8),

d Biomedical Analysis 145 (2017) 641–650

were purchased from Sigma Aldrich, France. 6-phenyl-1,3,5-triazine-2,4-diamine (BGA, CAS n◦ 91-76-9), 1,3-dicyclohexylurea(DCU, CAS n◦ 2387-23-7) were purchased from Alfa Aesar. Formicacid, >98% and ammonium hydroxide solution, 28.0-30.0% NH3basis were purchased from Sigma Aldrich.

The reference standards for biotest: 17�-estradiol (E2),5�-dihydrotestosterone (DHT), ICI 182,780 were supplied bySigma-Aldrich. Cell culture media were purchased from Gibco(Invitrogen), dextran coated charcoal stripped fetal bovine serum(DCC-FBS) and FBS were obtained from PAN Biotech.

Steady Glo Luciferase Assay reagent was purchased fromPromega. Clear bottom white luminometer 96-well plates for cellculture were supplied by Greiner Bio-One. Luciferase activity wasmeasured in a plate reader Chameleon (Hidex).

C18 (360 mg sorbent) SPE cartridges were obtained fromWaters. A Supelco 12-column vacuum manifold connected to avacuum pump was used for the SPE. Ultrapure deionized waterwas obtained from a Milli-Q RG, Millipore apparatus (Millipore SAS,France).

2.2. Cosmetic samples

Four aluminium coated flasks, containing 50 mL of cosmeticwater and corresponding to four suppliers (MC, AA, RR, BBF) wereprovided by Pierre Fabre laboratories. Each coated flask and a con-trol glass flask were filled with the same water solution and storedin two conditions: at ambient temperature or at 40 ◦C for threemonths to check the effect of storage. This protocol was arisen fromthe ASTM F1980-16 standard guide [9].

2.3. Extraction procedure

The extraction procedure is described on Fig. 1. Briefly extractionwas performed using SPE method concentrated by 8000-fold andanalysed by GC–MS and HPLC–MS, the same sample extract wasused after 250 times dilution to perform bioassays.

2.4. Solid phase extraction procedure

Sample extraction procedure has been optimized by Wagneret al. [10]. SepPack C18 cartridges was used for solid phase extrac-tion. Conditioning of the cartridges was performed with 2 × 4 mL ofacetone, followed by 2 × 4 mL of water. After loading of 400 mL ofeach sample at a flow-rate of 15 mL/min, the cartridges were driedfor 5 min by applying vacuum. The elution was done with 4 mL ofacetone, followed by 4 mL of methanol in glass vial containing 50 �LDMSO. Due to its high melting point, DMSO functions as a so-calledkeeper that retains volatile compounds during evaporation. Ace-tone and methanol were removed under a gentle stream of nitrogenat 40 ◦C yielding a final extract of 50 �L DMSO (concentration fac-tor 8000). SPE extracts were stored at −20 ◦C before their analysisin the in vitro study and by GC–MS/HPLC–MS. An extraction blankusing the same SPE extraction was performed on ultrapure water.In order to determine the hormonal activity, the SPE extracts werediluted 1, 2, 4, 8, 16, 32 and 64 times in DMSO. Each dilution wastested at a final concentration of 0.4% (v/v) in the culture medium.

2.5. Gas chromatography mass spectrometry analysis (GC–MS)

One �L of the 10 times diluted SPE extract was injected in asplitless mode of an Agilent 6890 series chromatograph. The tem-perature of the injector was 280 ◦C. A fused-silica capillary column

(30 m x 0.32 mm ID x 0.5 �m film thickness) coated with a station-ary phase DB-5MS (J&W Scientific, Folsom, CA, USA) was used.Helium was used as the carrier gas with a constant flow rate of1.4 mL/min and the oven temperature was programmed as follows:

E. Bou-Maroun et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650 643

and an

tfitMmsafiNdic

Lcluwtt

orofoc

2(

acWswfvf

Fig. 1. Extraction

he initial oven temperature was set at 150 ◦C for 1 min, it wasrstly increased to 250 ◦C at a rate of 20 ◦C/min and then increasedo 320 ◦C at a rate of 10 ◦C/min with a final isotherm of 5 min.

ass spectrometry was performed on a mass selective detectorodel 5973 (Agilent Technologies) operated on the electron ioni-

ation mode (70 eV). Ion source was set at 230 ◦C and transfer linet 300 ◦C. The solvent delay was 4 min. Compounds were identi-ed by comparison with mass spectra libraries (WILLEY138 andIST database). They were confirmed by the injection of pure stan-ards and by the calculation and comparison of the GC retention

ndex with those found in published data calculated under the sameonditions.

Quantification of BPA was done after extracting the ion m/z 213.inearity was obtained in the range of 0.65–13 mg/L. The correlationoefficient was R2 = 0.9991. The limit of detection (LOD) and theimit of quantification (LOQ) of the GC–MS method were calculatedsing the following equations: LOD = (3.3x SD/S); LOQ = (10x SD/S),here SD is the standard deviation of the GC–MS response and S is

he slope of the calibration curve. The LOD and the LOQ for BPA inhe SPE extract were respectively 0.45 mg/L and 1.36 mg/L.

Quantification of HMBT, BGA and DCU were determined basedn the total ion chromatogram (TIC). Linearity was obtained in theange of 89–298 mg/L for HMBT with a correlation coefficient R2

f 0.9469. Linearity was obtained in the range of 102.5–410 mg/Lor BGA with a correlation coefficient R2 of 0.9987. Linearity wasbtained in the range of 103–515 mg/L for DCU with a correlationoefficient R2 of 0.9729.

.6. Liquid chromatography mass spectrometry analysisHPLC–MS)

The SPE extract was diluted 10 times with the mobile phasend analytical standards were dissolved in the mobile phase at aoncentration of 100 �g/mL. Analyses were performed on a 2695

aters HPLC system which included an autoinjector and a thermo-tatic oven for the column. A Waters Acquity QDA mass detector

as used. Separations were performed on a Cortecs C18 column

rom Waters 2.7 �m (4.6 × 150 mm) at 28 ◦C. A mixture of two sol-ents were used as mobile phase: Solvent A (water + 0.1% v/v oformic acid for the positive ion mode or 0.1% v/v of ammonium

alysis procedure.

hydroxide for the negative ion mode) and solvent B (acetoni-trile + 0.1% v/v of formic acid or 0.1% of ammonium hydroxide). Thegradient used for the separation started at 10% of B and was main-tained for 2 min. It was increased linearly to 90% of B during 5 min,maintained at 90% of B for 2 min, finally decreased to 10% in 2 minand maintained for 2 min at 10% of B. The flow rate was 0.6 mL/min.The compounds were analysed using electrospray ionisation (pos-itive or negative mode). The mass spectrometer was scanned inthe 50–1200 m/z range. Analytical conditions were as follows: cap-illary voltage 0.8 kV, cone voltage 30 V, source block temperature120 ◦C. Injection volume was 5 �L. Analyses were performed induplicate.

2.7. Bioassay method

2.7.1. CellsFor the cell-based estrogen receptor (ER) mediated bioas-

say, stably-transfected hER�-HeLa-9903 cells were obtained fromthe Japanese Collection of Research Bioresources (JCRB N◦1318)cell bank. These cells contain stable expression constructs forhuman ER� and firefly luciferase, respectively. The latter is undertranscriptional control of five ERE promoter elements from thevitellogenin gene. Cells were maintained in Eagles Minimum Essen-tial Medium (EMEM) without phenol red, supplemented withkanamycin (60 mg/L) and 10% FBS (v/v), in an incubator under 5%CO2 at 37 ◦C. Upon reaching 75–90 % confluency, cells were sub-cultured two times prior to exposure to the material extracts.

For the cell-based androgen receptor (AR) mediated bioassay,MDA-kb2 cells derived from the MDA-MB-453 breast cancer cellline and stably transfected with the murine mammalian tumourvirus (MMTV)-luciferase.neo reporter gene construct [11] wereobtained from the ATCC (N◦ CRL-2713). Cells were routinely main-tained in Leibowitz-15 (L-15) medium supplemented with 10% FBS(v/v) in a humidified incubator at 37 ◦C without additional CO2.Cells were sub-cultured when confluent over a maximum of 10passages.

2.7.2. Transcriptional Activation AssaysThe assay for estrogenic activity was performed according to the

OCDE guideline TG455 as follows. Prior to experiments, Hela-9903

644 E. Bou-Maroun et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650

Fig. 2. Estrogenic activities recorded in the Hela-9903 reporter gene assay (OCDE guideline TG455). Four coded samples were stored at ambient temperature or at 40 ◦C fort oncenp standi 1 nM

c(acw

fPmmw1a

wwfippepascg

amaumccs

wimrea

hree months. For each material extracts, seven concentrations (expressed as CF: cercentage of the maximum luciferase activity obtained with the agonist reference

ndependent experiments and 3 wells/experiment). The straight-line highlights 10%

ells were maintained in culture medium supplemented with 10%v/v) of DCC-FBS for at least two media-changes. Cells were seededt a density of 1 × 104 cells per well in 100 �L of phenol red freeulture medium supplemented with 10% DCC-FBS in clear bottomhite luminometer 96-well plates and allowed to attach for 3 h.

A modified version of the original protocol [11] was employedor testing extracts for androgenic and anti-androgenic activities.rior to experiments, MDA-kb2 cells were maintained in L-15edium supplemented with 10% (v/v) DCC-FBS for at least twoedia-changes. Cells were seeded at a density of 1 × 104 cells perell in 100 �L of phenol red free L-15 medium supplemented with

0% DCC-FBS in clear bottom white luminometer 96-well platesnd allowed to attach for 24 h.

After incubation, 50 �L of a 3-fold dosing medium was added inells. Dosing medium consisted of culture medium supplementedith 10% DCC-FBS and containing the carrier solvent (DMSO, 0.4%nal solvent concentration, negative control) or E2 (17�-estradiol,ositive control in the ER assay) or DHT (dihydrotestosterone,ositive control in the AR assay) or the respective DMSO samplextract (for measurement of agonist activity) or the indicated sam-le extract plus DHT (for measurement of androgenic antagonistctivity). DMSO sample extracts were represented by a dilutioneries (7 different concentrations) of the tested samples and twoontrol extracts: a blank of extraction (ultrapure water) and a waterlass sample (cf. supplementary data, Fig. S1).

All treatments were tested in at least 3 replicates per plate,nd each extract was tested in at least two independent experi-ents. DHT or E2 at 1 nM was used as a positive control in the

gonist assay. DHT at 0.25 nM (non-saturating concentration) wassed as a control to establish a baseline for co-exposure experi-ents to screen for AR antagonism. E2 at 0.025 nM (non-saturating

oncentration) was used as a control to establish a baseline foro-exposure experiments to screen for ER antagonism (substancestudied alone).

After 24 h of incubation, the luciferase activity was determinedith Steady Glo assay reagent according to the manufacturer’s

nstructions. All luminescence values were corrected for the

ean of the solvent control and then related to the maximal

esponse of standard ligand which was set to 100% (E2max forstrogenicity or DHTmax for androgenicity or DHTsuboptimal for anti-ndrogenicity).

tration factor of initial water) were tested. RTA: Relative Transcriptional Activity:ard E2 (17�-estradiol) at 1 nM (=100%). Results are expressed as mean ± SD (n = 6, 2

E2 normalised RTA as a threshold for categorizing positive data.

2.7.3. CytotoxicityThe cells were visually examined under a phase-contrast micro-

scope for signs of cytotoxicity such as detachment, vacuolization,membrane degradation or growth inhibiting effects. Furthermore,luminescence readings significantly below (10%) the vehicle con-trol level (0%) in the agonist mode were considered cytoxic for thecells.

3. Results and discussion

3.1. Estrogenic and anti-androgenic activities of water extracts

In this study, we performed extraction of the aluminium coat-ing to reflect a potential release of unknown NIAS. This approachreflects the interaction that could occur between the material andthe packaged matrix as a worst-case scenario in term of exposure.Then, the first step consisted to check extract hormonal activitieswith bioassays. Except the sample BBF sold to be exempt of badge,all extracts induced an estrogenic activity in the human cell lineup to 30% compared to controls, which is significant according tothe OCDE 455 guideline (Fig. 2). The positive answer was observedat the highest concentration (32-fold concentrated). We demon-strated that this estrogenic activity was specific as the effect wasabolished using an ER antagonist such as ICI 182,780 (cf. supple-mentary data, Fig. S2). Then, aluminium coatings water extractsinduced positive answers in term of estrogenicity in bioassays, RRbeing the most potent (RR estrogenicity represented 160.2 pg EqE2/L) followed by the AA and MC which were similar in term ofpotency. Interestingly, we also demonstrated that a storage contactat 40 ◦C considered as accelerated conditions [9] could increase theestrogenic activity. Activities were positive only at 40 ◦C for MC andwere more pronounced for AA and for RR. None of extract inducedan androgenic effect (cf. supplementary data, Fig. S3).

BBF and MC were not anti-androgenic (data not shown). Incontrast, RR induced a significant concentration dependent anti-androgenic effect (60% at the highest concentration) without anycytotoxicity and the storage conditions did not impact the answer

(Fig. 3). We observed also with the AA extract, especially after 40 ◦C,a synergistic answer (130%) with DHT.

Using sensitive bioassays recommended in the level 2 for thescreening of potential ED [12], these data suggest the presence of

E. Bou-Maroun et al. / Journal of Pharmaceutical an

Fig. 3. Anti-androgenic activities recorded in the MDA-Kb2 reporter gene assay. AAand RR samples were stored at ambient temperature or at 40 ◦C for three months.For each material extracts, seven concentrations (expressed as CF: concentrationfactor of initial water) were tested.RTA: Relative Transcriptional Activity: percentage of the suboptimal luciferase activ-ity obtained with the agonist reference standard DHT (dihydrotestosterone) at0in

soeabsFfnsr

3

amif

pispIbabs

.25 nM (=100%). Results are expressed as mean ± SD (n = 6, 2 independent exper-ments and 3 wells/experiment). The straight-line highlights 70% 0.25 nM DHTormalised RTA as a threshold for categorizing positive data.

ubstances with agonist activity of estrogen receptor or/and antag-nist activity on androgen receptor in aluminium coatings waterxtracts. We can confirm that it is due to the aluminium coatingss no any positive answer was observed with aqueous solution andlanks (cf. supplementary data, Fig. S1). The relevance of bioas-ays is already recognized to assess hazard with extracts such asCM, they even allowed to discriminate non-food grade samplesrom food grade samples [8]. Moreover, the observed toxicity can-ot always be explained by the known and the identified migrantsuggesting that testing the whole migrate offers an opportunity toeduce uncertainty [4].

.2. Chemical analysis of the coating extracts

The chromatographic runs of extracts obtained after GC–MSnalysis are shown on Fig. 4. Samples RR and RR40 contained aaximum of chromatographic peaks which corresponds to a max-

mum of detected compounds (12). BBF and BBF40 extracts (badgeree samples) contained a minimum of detected compounds.

Compound n◦ 9 was present in all samples. However, this com-ound was not detected in a GC–MS column blank but detected

n ultrapure water extraction blank and in a water sample extracttored in glass. It was considered as a contamination product whichrobably came from the SPE syringe polymer or frits polymer.

ndeed, the cartridges consist of moulded polypropylene syringe

odies with two polyethylene frits. As Compound 9 appears, inll samples except MC40, in higher concentration than that in thelank: this contaminant could be better eluted in the presence ofamples.

d Biomedical Analysis 145 (2017) 641–650 645

Twelve compounds were detected by GC–MS in most of SPEextracts (cf. supplementary data, table S1). These 12 compoundswere tentatively identified by comparing their electron impactmass spectrum with mass spectra libraries (WILLEY138 and NISTdatabase). Table 1 gathers the identified compounds by GC–MSanalysis (8 substances). Only compounds with a high percentageof matching between experimental and database mass spectrumwere retained. To confirm the identity of these 8 compounds, acomparison between the calculated retention index (RI) and thereference RI found in the bibliography was done. All compoundswere identified by at least 3 of their fragment ions, reported in thecolumn m/z.

Compounds 1, 4, 9 and 11 could not be identified because ofthe low matching between their experimental mass spectrum withthe one of the database. This is probably due to co-elution. Chang-ing chromatographic conditions, such as reducing the temperatureoven ramp could solve this problem but it is time consuming.

Some substances were partly identified (2,3, 10, 12). In this case,we obtained a high score matching between their mass spectrumand the ones of the database and we confirmed their retentionindex with the ones of the bibliography except for compound10 for which no data was available. Among these substances,1,3-Benzothiazol-2(3H)-one (n◦ 2) was detected in MC and AAextract, we do not have any explanation about its presence but thissubstance is known to be a degradation product of 2-mercapto-benzothiazole (MBT), a rubber vulcanization accelerator in rubbermanufacture [13]. 2-(1-Phenylethyl) phenol (n◦ 3) was present inRR, MC and BBF extract. It is a structure quite similar to BPA, prob-ably a by-product of BPA. It has been found in raw chloroprenerubber extract [14]. 1,8-Diazacyclo-tetradecane-2,9-dione (n◦ 10),a dimer of �-caprolactam appeared present in every samples. It hasbeen detected by GC–MS in a polyamide resin extract [15]. The com-mercially important engineering polymer polyamide-6 (PA-6), alsoknown as nylon-6, is produced from the monomer �-caprolactam.Furthermore, (9Z)-9-Octadecenamide (n◦ 12) was also present inall coating extracts. It is a slip-agent for extrusion of polyethylene,known as a wax, ink and lubricating oil additive. It is also used as aslip and antiblock agent for plastics-ESP polyethylene film.

Substances N◦ 5, 6, 7, 8 were identified with high certainty byinjection of the corresponding pure standards in GC–MS and con-firmation of their mass spectrum and their retention index. Thesesubstances are respectively HMBT, BGA, DCU and BPA. Moreover,the SPE extracts were analysed by HPLC–MS. HMBT, BGA, DCU andBPA were also confirmed in HPLC–MS by comparing their retentiontime with analytical pure standards and by checking their [M+H]+

or [M-H]− ion mass values. HMBT was detected in positive ion modeat 9.03 min, its mass spectrum showed the ion [M+H]+ = 212.1.BGA was detected in positive ion mode at 5.31 min, its massspectrum showed the ion [M+H]+ = 188.2. DCU was detected inpositive ion mode at 9.69 min, its mass spectrum showed the ion[M+H]+ = 225.4. BPA was detected in negative ion mode at 8.67 min,its mass spectrum showed the ion [M−H]− = 227.0 (cf. supplemen-tary data, Fig. S4). The double confirmation of HMBT, BGA, DCU andBPA by GC–MS and HPLC–MS constitute a high certainty of identi-fication. Because the two methods are conceptually different fromone another, both methods would not be subject to the same inter-ferences. Then, both methods could be used to confirm each other[16].

Then, HMBT, BGA, DCU and BPA were quantified by GC–MSusing external standard calibration. HMBT was identified in all alu-minium coating extracts. HMBT is a benzothiazole derivative, it hasbeen identified as a NIAS in syringes and formed from MBT and

ethylene oxide during sterilization [17]. It has been also identifiedas a micropollutant in water and as leachate from tire [18]. Its con-centration in water samples (cf. Fig. 1) varied between 145.8 �g/L

646 E. Bou-Maroun et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650

atogr

fH

tAai

Fig. 4. GC–MS chrom

or BBF40 and 303.6 �g/L for MC40. The lowest concentration ofMBT corresponds to the badge free sample (BBF40).

BGA was only detected in AA and AA40 samples, its concentra-

ions in water samples were 239.5 �g/L for AA and 264.9 �g/L forA40. BGA, is a well-known additive, it is used as a cross linkinggent in PVC and contributes to a variety of PVC coatings improv-ng hardness, gloss and resistance and shelf life. It has been already

ams of SPE extracts.

identified as a released compound after migration from PTFE withFCM [19]. BGA is listed in plastic FCM regulation with a SML of5 mg/kg of food (Commission Regulation (EU) No 10/2011) and also

in the US FDA CFR (2014). It has been mentioned that BGA migrationwas depending of the temperature storage and increased duringlong storage [20]. Our data in this study are in accordance as its

E. Bou-Maroun et al. / Journal of Pharmaceutical an

Tab

le

1C

omp

oun

ds

iden

tifi

ed

by

GC

–MS

in

SPE

sam

ple

extr

acts

.

Com

pou

nd

Nu

mbe

rIU

PAC

nam

e

CA

S

nu

mbe

r

Form

ula

%

dat

abas

em

atch

inga

Cal

cula

ted

RI

Ref

eren

ce

RI

Iden

tifi

cati

on

m/z

iden

tifi

edio

ns

2

1,3-

Ben

zoth

iazo

l-2(

3H)-

one

934-

34-9

C7H

5N

OS

87

1723

1716

[28]

++

151,

123,

963

2-(1

-Ph

enyl

eth

yl)

ph

enol

4237

-44-

9

C14

H14

O

86

1755

1721

[29]

++

198,

183,

165

5

2-(1

,3-B

enzo

thia

zol-

2-yl

sulf

anyl

)eth

anol

(HM

BT)

4665

-63-

8

C9H

9N

OS 2

99

1989

1890

[30]

+++

211,

181,

167

6

6-Ph

enyl

-1,3

,5-t

riaz

ine-

2,4-

dia

min

e

(BG

A)

91-7

6-9

C9H

9N

598

2064

1972

[30]

+++

187,

144,

103

7

1,3-

Dic

yclo

hex

ylu

rea

(DC

U)

2387

-23-

7

C13

H24

N2O

95

2097

2075

[30]

+++

224,

143,

998

4,4′ -

(2,2

-Pro

pan

ediy

l)d

iph

enol

(BPA

)

80-0

5-7

C15

H16

O2

95

2221

2185

[30]

+++

228,

213,

119

10

1,8-

Dia

zacy

clo-

tetr

adec

ane-

2,9-

dio

ne

5776

-79-

4

C12

H22

N2O

297

2334

N.A

.

+

226,

198,

112

12

(9Z)

-9-O

ctad

ecen

amid

e

301-

02-0

C18

H35

NO

89

2419

2397

[30]

++

281,

264,

140

a:

This

per

cen

tage

corr

esp

ond

s

to

the

sim

ilar

ity

betw

een

the

exp

erim

enta

l mas

s

spec

tru

m

and

the

one

of

the

dat

a

base

.+:

Iden

tifi

cati

on

by

com

par

ison

wit

h

mas

s

spec

tru

m

dat

abas

e

and

hig

h

scor

e

mat

chin

g.++

:

Iden

tifi

cati

on

by

com

par

ison

wit

h

mas

s

spec

tru

m

dat

abas

e

wit

h

hig

h

scor

e

mat

chin

g

and

by

com

par

ison

wit

h

rete

nti

on

ind

ex

dat

abas

e.++

+:

Iden

tifi

cati

on

by

inje

ctio

n

of

pu

re

stan

dar

ds

and

con

firm

atio

n

of

mas

s

spec

tru

m

and

rete

nti

on

ind

ex.

N.A

. =

Not

avai

labl

e.

d Biomedical Analysis 145 (2017) 641–650 647

concentration tended to increase in the AA extract after 40 ◦C (8.5instead of 7.7 �g/L).

DCU was present in all extracts except RR and RR40. It waspresent only after 40 ◦C in the BBF extract. DCU represented themost concentrated substance among the substances identified withcertainty, the highest concentration was recovered in MC and AAextracts. Its concentration in water extracts was 221.4 �g/L forBBF40 and two-fold higher (444.5 �g/L) for AA. DCU was identifiedas releasing chemical from polyurethane adhesives in food contactmaterials [21].

As expected, BPA was not found in badge free samples BBF andBBF40. It was only detected in RR samples; its concentration was0.9 �g/L for RR and increased up to 7.5 �g/L for RR40 suggest-ing a temperature effect on BPA release. The fact that more BPAis released with time and temperature is not surprising and inaccordance with the literature, as it has been already shown withepoxy-based resins and polycarbonate [22].

3.3. Hormonal activities of identified substances in the waterextracts

Four substances were identified with high certainty, in the alu-minium coated extracts, by GC–MS and HPLC–MS: HMBT, BGA,DCU and BPA (a well-known endocrine disruptor). We decidedto check the agonist or antagonist hormonal activities of the sin-gle substances (when commercially available) in our experimentalconditions to verify if the positive answer observed in extractscould be due to the presence of a specific chemical.

3.3.1. Estrogenic activitiesBPA as expected was estrogenic at 0.114 mg/L and up to

2.28 mg/L. The BPA concentration in RR and RR40 water sampleswas also determined in the bioassay model. A calibration curve, ERactivity versus log (BPA concentration), obtained from Fig. 5a wasused. The concentration of BPA determined in the bioassay methodwas then 3.8 �g/L for RR and 4.5 �g/L for RR 40.

DCU, a carbanilide analogue of triclosan was the most activeestrogenic chemical in this study (behind BPA) with a concentra-tion dependency starting at 2.46 mg/L and up to 7.4 mg/L (Fig. 5a).However, at the highest tested concentration, a precipitate wasobserved indicating a limit of DCU solubility. DCU acted also withsynergism with E2 (the positive reference at 0.025 nM) at a con-centration starting at 0.74 and up to 7.4 mg/L (Fig. 5b). Our data arein accordance with the study of Ahn et al. [21], showing that DCUinduced ER dependant gene expression at 2.24 mg/L and inducedby 2.5-fold E2 activity using reporter gene bioassays.

BGA was weakly estrogenic with a concentration dependencyeffect at 18.7 mg/L and up to 46.8 mg/L. Higher concentrations werecytotoxic for the cells. BGA acts with synergism at 1.87 and athigher concentrations in the presence of E2 (Fig. 5b). This studydemonstrates for the first time that BGA is a weak but biologicallysignificant estrogenic chemical in vitro.

HMBT was only estrogenic at the highest concentration(21.13 mg/L) with an activity in the same order than BGA (Fig. 5a).HMBT has been shown to be agonist on nuclear receptors suchas AhR (Aryl hydrocarbon Receptor, a ligand dependent transcrip-tion factor). Indeed, analysis of crude extracts revealed that mostextracts contained substantial AhR agonist activity using bioassays(CALUX cell lines). Tire extracts fractionations were also able toinduce P450 1A enzymatic activity mediated by the AhR [23]. HMBT

increased the estrogenic activity of E2 at the highest concentra-tion (Fig. 5b). Regards to synergism, this data is in accordance witha study on benzothiazoles derivatives extracts showing that theywere able to enhance AhR dependent gene expressions of TCDD

648 E. Bou-Maroun et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650

Fig. 5. Estrogenic (a) and anti-estrogenic (b) activities of substances identified with certainty in extracts in the Hela-9903 reporter gene assays. RTA: Relative TranscriptionalA tandarP ssed a4 GA is

(A

3

daapa

(aFc1

isas

3p

actte

to(o

ctivity: percentage of the luciferase activity obtained with the agonist reference s: observed precipitation of substance, T: observed cytotoxicity. Results are expre,4′-(2,2-propandiyl)diphenol. HMBT is 2-(1,3-Benzothiazol-2-ylsulfanyl)ethanol. B

2,3,7,8-tetrachlorodibenzo-p-dioxin) (the most potent ligand ofhR) with synergism [24].

.3.2. Anti-androgenic activitiesDCU, BGA or HMBT were not androgenic in experimental con-

itions (cf. supplementary data, Fig. S5). However, DCU and BGActed with a weak but biological significant synergism with DHTt 0.25 nM. DCU may sensitize the complex receptors associatingroteins similar to cofactors or co-activators in cells containing ARs it has been already suggesting with carbanilides [25].

HMBT was the only anti-androgenic at high concentrations21.13 mg/L) but at the highest concentration, this effect was notnymore specific due to a cytotoxic effect (cf. supplementary data,ig. S5). As expected, BPA was anti-androgenic in our experimentalonditions but compared to HMBT at lower concentration (around00-fold lower).

Taking together, then these data demonstrated hormonal activ-ties of the NIAS identified with certainty, synergism observeduggest that some of them like BGA, DCU could increase hormonalctivities if active substances are also present at low concentrationsuggesting that “cocktail effect” could occur in such extract.

.4. Combination of bioassays and analytical methods to detectotential endocrine substances in the water extract

Combining the bioassay’s answers with the chemical analysesnd after a quantification of the active substances by GC–MS, theoncentration of each biologically active substance was comparedo its respective activity threshold (Table 2). The objective waso identify in aluminium coated extracts which substance couldxplain the hormonal activities observed in the bioassay.

Regards to estrogenic activities (Table 2), RR extract induced

he most important estrogenic agonist answer with the presencef BPA. Then the effect could be due to BPA as its concentration0.03 mg/L) was enough high to be detected by the bioassay (thresh-ld of 0.02 mg/L). In contrast, HMBT concentration (around 7 mg/L)

d E2 at 1 or 0.025 nM (=100%).s mean ± SD. n = 12 (2 independent experiments and 6 wells/experiment). BPA is

6-phenyl-1,3,5-triazine-2,4-diamine and DCU is 1,3-dicyclohexylurea.

was under its biological threshold (21.1 mg/L, 96-fold higher com-pared to BPA). Furthermore, after storage conditions at 40 ◦C, theestrogenic activity of the RR40 extract was higher and in paral-lel, the BPA concentration was also higher (8-fold higher than atambient temperature). Furthermore, it is important to note thatthe concentrations of BPA in water samples calculated from theGC–MS method (0.9 �g/L for RR and 7.5 �g/L for RR 40) were inthe same order of magnitude when calculated with the bioassaymethod (3.8 �g/L for RR and 4.5 �g/L for RR 40) suggesting a goodsensitivity of bioassay.

For MC and AA, the observed hormonal activities could be dueto DCU as DCU was the most concentrated substance (7.2 up to14.2 mg/L) with the lowest biological threshold (0.74 mg/L).

BGA was only present in AA and AA40 extracts and its concentra-tion always below its biological activity threshold which was muchhigher than DCU threshold, suggesting that BGA was not involvedas single substance in estrogenic activities observed in AA extract.

Furthermore, even if DCU was detected in BBF40 and at a con-centration similar to MC40, the estrogenic activity was not positivein the BBF40 extract. In parallel, a precipitate was observed in theculture media (limit of solubility) which could explain the absenceof answer.

Regards to HMBT, present in all extracts whatever the tempera-ture is, its concentration was always below its biological threshold,which was the highest threshold (21.1 mg/L) determined in thisstudy. RR extracts were the only samples which induced an anti-androgenic answer (Table 2), two compounds (HMBT and BPA)were active when tested alone in the bioassay: HMBT being a weakanti-androgenic compound as its biological threshold (21.1 mg/L)was 95-fold higher compared to BPA. However, none of the sub-stance alone could explain the global positive activity due to theirconcentrations below their respective biological threshold in the RRextract (Table 2). The positive anti-androgenic activity could be due

to a synergistic effect between HMBT and BPA. Indeed, we observeda synergism between HMBT and DHT, the same effect may occurbetween BPA and HMBT in the extract conditions. In RR40 extract,

E. Bou-Maroun et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 641–650 649

Table 2Summary of the estrogenic and of the anti-androgenic activities in the bioassay.

Estrogenic activity Anti-androgenic activity

Sample Identified activesubstance

Active substanceconcentration inthe bioassay(mg/L)

Activity of theSPE extract

Activitythreshold (mg/L)

Observation Activity of theSPE extract

Activitythreshold (mg/L)

Observation

MC HMBT 6.8 +/− 21.1 −DCU 13.3 0.74

√MC40 HMBT 9.7 + 21.1 −

DCU 7.2 0.74√

AA HMBT 6.6 +/− 21.1 −DCU 14.2 0.74

√BGA 7.7 18.7

AA40 HMBT 4.9 + 21.1 −DCU 8.2 0.74

√BGA 8.5 18.7

BBF HMBT 6 − 21.1 −BBF40 HMBT 4.7 − 21.1 −

DCU 7.1 0.74RR HMBT 9.2 + 21.1 + 21.1

BPA 0.03 0.02√

0.22RR40 HMBT 7.4 + 21.1 + 21.1

BPA 0.24 0.02√

0.22√

(( e activ

iatismsrsmeaocgcaErmbs

4

aat4tDwDbcpb(cp

+) = high activity, (+/−) = medium activity and (−) = low activity.√) means that the active substance concentration in the bioassay is higher than th

n contrast, as the BPA concentration was 8-fold higher, its presencet this concentration could explain the positive answer obtained inhe extract. Concerning the mixture effect, determining the activ-ty of the extract is more realistic as not only one chemical buteveral are released by the packaging. Although chemicals haveinor endocrine activity on their own, they are able to enhance

teroids activity in vitro. We demonstrated indeed synergistic effectegards to estrogenicity and anti-androgenicity with DCU and BGA,uggesting that in a mixture, BGA or DCU could increase a low hor-onal activity underlying that it makes sense to check packaging

xtract with all released chemicals, as some endocrine disruptorsct at very low concentrations and with cocktail effect [6]. Thisbservation points out that it is important to verify the extract inomplement to the chemistry. Concerning consumer exposure, theenerally accepted figure is 17.4 g of cosmetics per day [26]. Wean conclude from this study that the risk assessment is negligibles the most responsive extract represented 160.2 pg Equivalent ofstradiol (EQ) activity/L. Converting to the mass of cosmetic, it willepresent 2.78 pg EQ of estradiol per day, which is very low anduch lower (10 fold) than in bottled mineral waters using yeasts

ioassay [27] but higher (13 fold) than in the Wagner and Oehlmanntudy [10] using the E-Screen assay.

. Conclusion

Four extracts of aluminium coatings were tested for hormonalctivities using gene reporter bioassays, 3 were weakly estrogenicnd the most active was both estrogenic and anti-androgenic,his effect being probably due to the BPA presence. Furthermore0 ◦C increased extract estrogenic activity. Using GC–MS analysis,hree chemicals besides BPA were identified with certainty: BGA,CU, and HMBT. They were also confirmed by HPLC–MS. Theyere considered as estrogenic NIAS with the following potencyCU > BGA > HMBT. Only HMBT was weakly anti-androgenic. Com-ining both methodologies, we demonstrated that bioassays andhemical analysis were complementary methods to identify NIASresence which could induce hormonal activities in extracts. Using

ioassays, we also observed synergistic effects of identified NIASbut not predicted) such as BGA and DCU. Then, this strategyombining chemical analysis and bioassays permits to monitorackaging intended to be in contact with cosmetics, to protect

[

ity threshold.

consumer safety with certainty, (especially pregnant women andinfants) in term of hormonal chemical activities and to increase theconsumer confidence.

Acknowledgment

This research study was funded by Pierre Fabre Dermo-Cosmétique.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jpba.2017.07.061.

References

[1] K. Grob, M. Biedermann, E. Scherbaum, M. Roth, K. Rieger, Foodcontamination with organic materials in perspective: packaging materials asthe largest and least controlled source? a view focusing on the Europeansituation, Crit. Rev. Food Sci. Nutr. 46 (7) (2006) 529–535.

[2] K. Grob, P. Camus, N. Gontard, H. Hoellinger, C. Joly, A.C. Macherey, D. Masset,F. Nesslany, J.F. Régnier, A.M. Riquet, P. Saillard, D. Ribera, Need for a bettersafety evaluation of food contact materials produced from resins, FoodControl 21 (5) (2010) 763–769.

[3] C. Nerin, P. Alfaro, M. Aznar, C. Domeno, The challenge of identifyingnon-intentionally added substances from food packaging materials: a review,Anal. Chim. Acta 775 (2013) 14–24.

[4] J. Muncke, Endocrine disrupting chemicals and other substances of concern infood contact materials: an updated review of exposure, effect and riskassessment, J. Steroid Biochem. Mol. Biol. 127 (1–2) (2011) 118–127.

[5] E.R. Kabir, M.S. Rahman, I. Rahman, A review on endocrine disruptors andtheir possible impacts on human health, Environ. Toxicol. Pharmacol. 40 (1)(2015) 241–258.

[6] E. Silva, N. Rajapakse, A. Kortenkamp, Something from nothing − eight weakestrogenic chemicals combined at concentrations below NOECs producesignificant mixture effects, Environ. Sci. Technol. 36 (8) (2002) 1751–1756.

[7] S. Koster, M.H. Bani-Estivals, M. Bonuomo, E. Bradley, M.C. Chagnon, M.L.Garcia, F. Godts, T. Gude, R. Helling, P. Paseiro-Losada, G. Pieper, M. Rennen, T.Simat, L. Spack, Guidance on Best Practices on the Risk Assessment of NonIntentionally Added Substances (NIAS) in Food Contact Materials and Articles,ILSI Europe, Brussels, 2015, 68 pp.

[8] I. Severin, E. Souton, L. Dahbi, M.C. Chagnon, Use of bioassays to assess hazardof food contact material extracts: state of the art, Food Chem. Toxicol. 105(2017) 429–447.

[9] ASTM F1980-16, Standard Guide for Accelerated Aging of Sterile BarrierSystems for Medical Devices, in: W.C. ASTM International, PA, (Ed.) 2016.

10] M. Wagner, J. Oehlmann, Endocrine disruptors in bottled mineral water:estrogenic activity in the E-screen, J. Steroid Biochem. Mol. Biol. 127 (1–2)(2011) 128–135.

6 tical an

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

50 E. Bou-Maroun et al. / Journal of Pharmaceu

11] V.S. Wilson, K. Bobseine, C.R. Lambright, J.L.E. Gray, A novel cell line,MDA-kb2, that stably expresses an androgen- and glucocorticoid-responsivereporter for the detection of hormone receptor agonists and antagonists,Toxicol. Sci. 66 (1) (2002) 69–81.

12] S. Satya, M. Wade, U. Hass, H. Holbech, M. Lofstedt, A.M. Vinggaard, K.H. Tyle,P. Juul Nielsen, M.L. Holmer, S. Christiansen, Guidance Document onStandardised Test Guidelines for Evaluating Chemicals for EndocrineDisruption, Organisation for Economic Cooporation and Development (OECD),ENV/JM/MONO, 2012, pp. 22.

13] N-Cyclohexylbenzothiazol-2-sulphenamide, CAS No: 95-33-0, Summary RiskAssessment Report, European Communities, European Chemicals Bureau,ECB, 2008, 2017.

14] N.H. Nilsson, V. Pedersen, Mapping and release of chemical substances fromproducts made of chloroprene, Survey of Chemical Substances in ConsumerProducts, No. 51, Danish Ministry of the environment, EnvironmentalProtection Agency, 2004.

15] T. Rothenbacher, Improvements in the Analysis of Food ContaminationsDeriving from Packaging Materials, Fakultät Naturwissenschaften UniversitätHohenheim, Ph.D. Thesis, p. 142, 2009.

16] L.S. De Jager, G.A. Perfetti, G.W. Diachenko, Comparison ofheadspace-SPME-GC–MS and LC–MS for the detection and quantification ofcoumarin, vanillin, and ethyl vanillin in vanilla extract products, Food Chem.107 (4) (2008) 1701–1709.

17] B.S. Middleditch, Journal of Chromatography Library – Volume 44. AnalyticalArtifacts. GC, MS, HPLC, TLC and PC, Elsevier Science, Amsterdam, 1989 (p.423).

18] C. Hug, N. Ulrich, T. Schulze, W. Brack, M. Krauss, Identification of novelmicropollutants in wastewater by a combination of suspect and nontargetscreening, Environ. Pollut. 184 (2014) 25–32.

19] E.L. Bradley, W.A. Read, L. Castle, Investigation into the migration potential ofcoating materials from cookware products, Food Addit. Contam. 24 (3) (2007)326–335.

20] M. Vaclavikova, R. Paseiro-Cerrato, L. Vaclavik, G.O. Noonan, J. DeVries, T.H.Begley, Target and non-target analysis of migrants from PVC-coated cans

[

[

d Biomedical Analysis 145 (2017) 641–650

using UHPLC-Q-Orbitrap MS: evaluation of long-term migration testing, FoodAddit. Contam. A 33 (2) (2016) A352–A363.

21] K.C. Ahn, B. Zhao, J. Chen, G. Cherednichenko, E. Sanmarti, M.S. Denison, B.Lasley, I.N. Pessah, D. Kültz, D.P.Y. Chang, S.J. Gee, B.D. Hammock, In vitrobiologic activities of the antimicrobials triclocarban, its analogs, and triclosanin bioassay screens: receptor-based bioassay screens, Environ. HealthPerspect. 116 (9) (2008) 1203–1210.

22] J.E. Cooper, E.L. Kendig, S.M. Belcher, Assessment of bisphenol A released fromreusable plastic, aluminium and stainless steel water bottles, Chemosphere85 (6) (2011) 943–947.

23] G. He, T. Tsutsumi, B. Zhao, D.S. Baston, J. Zhao, S. Heath-Pagliuso, M.S.Denison, Third-generation ah receptor-responsive luciferase reporterplasmids: amplification of dioxin-Responsive elements dramatically increasesCALUX bioassay sensitivity and responsiveness, Toxicol. Sci. 123 (2) (2011)511–522.

24] D.S. Baston, M.S. Denison, Considerations for potency equivalent calculationsin the Ah receptor-based CALUX bioassay: normalization of superinductionresults for improved sample potency estimation, Talanta 83 (5) (2011)1415–1421.

25] C. -y. Chang, D.P. McDonnell, Androgen receptor-cofactor interactions astargets for new drug discovery, Trends Pharmacol. Sci. 26 (5) (2005) 225–228.

26] Notes of Guidance for testing of cosmetic substances and their safetyevaluation, 8th Revision 2012, Scientific Committee on Consumer SafetySCCS/1501/12, 2012.

27] B. Pinto, D. Reali, Screening of estrogen-like activity of mineral water stored inPET bottles, Int. J. Hyg. Environ. Health 212 (2) (2009) 228–232.

28] M. Grung, A. Kringstad, K. Baek, I.J. Allan, K.V. Thomas, S. Meland, S.B.Ranneklev, Identification of non-regulated polycyclic aromatic compoundsand other markers of urban pollution in road tunnel particulate matter, J.

Hazard. Mater. A 323 (2017) 36–44.

29] NIST Chemistry WebBook, 2017, http://webbook.nist.gov/chemistry/.(Accessed January 2017).

30] ChemSpider Search and Share Chemistry, 2017, http://www.chemspider.com/. (Accessed January 2017).