Sc

Na

b

a

ARRAA

KCFCLMICGCT

1

tcatbiatob[

fef

(

0d

Journal of Chromatography A, 1216 (2009) 4703–4712

Contents lists available at ScienceDirect

Journal of Chromatography A

journa l homepage: www.e lsev ier .com/ locate /chroma

equential fractionation with concurrent chemical and toxicologicalharacterization of the combustion products of chlorogenic acid

avneet Kaura, Martine Lacasseb, Alexandra Fürtösa, Karen C. Waldrona,∗, André Morinb,∗∗

Department of Chemistry, University of Montréal, C.P. 6128, succ. Centre-Ville, Montréal, Québec H3C 3J7, CanadaImperial Tobacco Canada Ltd., 3711, rue Saint-Antoine Ouest, Montréal, Québec H4C 3P6, Canada

r t i c l e i n f o

rticle history:eceived 20 January 2009eceived in revised form 30 March 2009ccepted 2 April 2009vailable online 9 April 2009

eywords:hlorogenic acidractionation

a b s t r a c t

Chlorogenic acid is the most abundant polyphenol found in the tobacco plant. The biological effects ofits combustion products remain largely unknown. In this study, chlorogenic acid was burned at 640 ◦Cfor 2 min and the particulate matter of the smoke was collected onto Cambridge filter pads followed byselective extraction in five different solvents. Various fractions of the chlorogenic acid combustion prod-ucts were tested for induction of micronuclei in V79 Chinese hamster fibroblast cells. Over 40 compoundswere identified in the dimethyl sulfoxide (DMSO) extract by high-performance liquid chromatographycoupled to electrospray time-of-flight mass spectrometry (HPLC/TOF-MS). The DMSO extract was thenfractionated into three major fractions by preparative LC. The fraction inducing the highest degree of

ombustioniquid chromatographyass spectrometry

n vitro micronucleus testatecholenotoxicity

toxicity was further separated into four sub-fractions. The sub-fraction responsible for the most toxicresponse was determined to contain catechol as its major component. The overall reproducibility of thecombustion, the extraction procedure and the chemical characterization of the compounds responsiblefor the toxicity in the chlorogenic acid smoke were evaluated by LC/TOF-MS.

© 2009 Elsevier B.V. All rights reserved.

ell proliferationobacco

. Introduction

Tobacco consists of over 2000 components and upon combus-ion generates more than 7000 compounds [1]. Due to the highlyomplex nature of tobacco smoke, the exact mechanisms of toxicityre still unknown. For instance, a number of lists of cigarette smokeoxicants have been published in recent years, some of which haveegun to estimate the relative toxicity of the compounds found

n tobacco smoke [2]. However, these approaches are unable toccount for the complex chemical profile and potential interactionshat may occur in cigarette smoke. Many studies have been carriedut on whole tobacco smoke in efforts to determine the correlationetween tobacco smoke components and their biological effects3–5].

An alternative approach is to study the individual componentsound in leaf tobacco, which upon combustion generate a vari-ty of bioactive species. Among the major groups of constituentsound in tobacco, the polyphenol group accounts for about 10% of

∗ Corresponding author. Tel.: +1 514 343 6516; fax: +1 514 343 7586.∗∗ Corresponding author. Tel.: +1 514 932 6161x2666; fax: +1 514 932 6882.

E-mail addresses: karen.waldron@umontreal.ca (K.C. Waldron), amorin@itl.caA. Morin).

021-9673/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2009.04.007

the leaf dry weight [6,7]. Among the polyphenols, chlorogenic acid(CGA) (3-[[3-(3,4-dihydroxyphenyl)-1-oxo-2-propenyl]oxy]-1,4,5-trihydroxycyclohexanecarboxylic acid) is the most abundant singleconstituent. It represents about 1–3% or higher of leaf dry weightof the tobacco plant [8,9]. Several studies have identified compo-nents found in smoke from the combustion of CGA [10–13], andother studies have identified CGA as well as some of its combustionproducts as being genotoxic and carcinogenic [14–16]. Combus-tion of CGA principally generates pyrocatechol (more commonlyknown as catechol), phenol, hydroquinone, quinide, benzene andbenzoic acid. Some of these phenolic compounds were reported tobe toxic [13,17,18] whereas catechol and phenol were reported toenhance carcinogenic processes induced by other compounds suchas polynuclear aromatic hydrocarbons [19].

A few groups have reported toxicological data on individualcomponents found in tobacco smoke [15,17,20–22]. A previousstudy [21] indicated that of twelve tobacco components tested, thecombustion products of the two polyphenols, CGA and lignin con-tained the most bioactive components, evaluated by the in vitro

micronucleus test (IVMNT). The IVMNT is an in vitro genotoxic-ity test used to identify chemicals that induce the formation ofsmall, membrane-bound deoxyribonucleic acid fragments, calledmicronuclei, in the cytoplasm of interphase mammalian cells[23–26]. CGA is the least complex and most readily available of

4 togr. A

tfctctttpaefc

2

2

dMc((5fhaapsr((eWpwiSe(stacis

2

mot2tt(t

2p

f

704 N. Kaur et al. / J. Chroma

he two above polyphenolic compounds found in tobacco, there-ore, it was chosen for further investigation. The objective of theurrent study was to identify the toxic compounds resulting fromhe combustion of CGA under atmospheric conditions. The so-alled incomplete combustion conditions were adopted to mimiche process occurring during the burning of a cigarette. A strategyo partially combust, extract, fractionate and concurrently evaluatehe chemical composition and relative toxicity of the combustionroducts of CGA by in vitro toxicological assays was designed. Ourpproach combines analytical chemistry and in vitro toxicology toxpand knowledge on the toxicity of smoke constituents generatedrom the partial combustion (i.e., burning) of one single tobaccoomponent, chlorogenic acid.

. Experimental

.1. Chemicals and reagents

All standards and reagents used for the combustion repro-ucibility study were supplied by Sigma–Aldrich (St. Louis,O, USA) and were of ≥99.0% purity unless otherwise indi-

ated: hydroquinone (CAS 123-31-9), phenol (108-95-2), m-cresol108-39-4), p-cresol (106-44-5), o-cresol (95-48-7), pyrocatechol120-80-9), resorcinol (180-46-3), 3,4-dihydroxybenzoic acid (90-0-3), caffeic acid (331-39-5), trans-cinnamic acid (140-10-3),erulic acid (1135-24-6), 2,5-dihydroxybenzoic acid (490-79-9), p-ydroxybenzoic acid (99-96-7), 1,2-cyclohexanedione (765-87-7)t 97%, p-coumaric acid (501-98-4) at 98% and CGA (327-97-9)t ≥95% purity. Glass wool (Pyrex brand wool filtering fiber) wasurchased from Corning (Big Flats, NY, USA). The HPLC gradeolvents used for the filter extraction and the CGA combustioneproducibility study were dimethyl sulfoxide (DMSO), methanolMeOH), dichloromethane (DCM), ethyl acetate (EA), acetonitrileACN), acetic acid and formic acid, all purchased from Fisher Sci-ntific (Whitby, Canada) and used without further purification.ater used for the filter extraction was either distilled water

urified using a Milli-Q system (Millipore, Billerica, MA, USA),hich consisted of a carbon cartridge, two high-capacity mixed

on exchange cartridges and a 0.45 �m filter (Chromatographicpecialties, Brockville, Canada) or HPLC grade water from Fisher Sci-ntific. Formic acid for HPLC/MS studies was obtained from FlukaBuchs, Switzerland). Benzoic acid (65-85-0) at 99.5% purity wasupplied by Laboratoire MAT (Beauport, Canada). Appropriate ven-ilation measures and protection of researchers were employed forll manipulations that involved the use of organic solvents andompounds known or suspected to be toxic. The operation of allnstruments used in this study was carried out according to theafety procedures recommended by the manufacturers.

.2. Sample preparation

Aliquots composed of 0.5 g CGA dissolved in 5 mL of MeOH wereixed with the aid of a vortex then deposited onto a matrix of 0.5 g

f glass wool in individual Petri dishes. To evaporate the MeOH,he sample was stored for at least 72 h in a conditioned room at2.5 ◦C with 60% relative humidity. Following the storage period,he corrected mass of CGA adsorbed on the matrix was determinedo ±1.0 mg by subtracting the glass wool matrix and Petri dish massincludes CGA adsorbed onto the Petri dish) from the total mass ofhe sample (mass of matrix, CGA aliquot and Petri dish).

.3. Partial combustion of CGA and collection/extraction of thearticulate phase

The CGA sample adsorbed onto the glass wool matrix was trans-erred from the Petri dish and packed (7.5 cm bed length) into a

1216 (2009) 4703–4712

quartz combustion tube (outer dimensions: 26.5 cm × 1.2 cm, wallthickness: 1 mm). A John Payne Tar Predictor (JPTP) (John PayneMachinery Spares, Winchester, UK) apparatus was used to burnCGA and collect the particulate phase of its smoke. The quartz tubethat contained the CGA sample was automatically driven into thefurnace where the burning process was conducted at 640 ± 10 ◦Cfor 2 min. During this time, atmospheric air was drawn throughthe quartz tube at 1.8 L/min, forming smoke that passed throughthe Cambridge filter of diameter 55 mm (Borgwaldt, Richmond, VA,USA) which trapped the particulate phase of the smoke, or total par-ticulate matter (TPM). Full combustion in air, by definition, shouldrender all organic compounds to CO, CO2 and H2O. Therefore, to beaccurate, the burning process employed here results in partial orincomplete combustion of CGA. Silicone grease was used to avoidleaking of smoke from the tubing at specific locations.

To allow for deposition of the particulate phase, each Cambridgefilter was set aside for a period of 15–60 min. The Cambridge filterwas weighed to ±0.1 mg before and after the burning process todetermine the mass of collected TPM. The particulate matter col-lected on the Cambridge filter was extracted under vacuum, usinga Büchner funnel, by adding drop-wise a specific volume of sol-vent as follows. For DMSO extraction, the volume of DMSO usedwas that needed to obtain a final concentration of 15 mg/mL ofTPM, assuming 100% extraction efficiency. For the other solvents,the extraction volume was fixed at 10 mL per filter to obtain asuitable volume for the subsequent biological assay. The extrac-tion solvent was then evaporated using a rotary evaporator (exceptwhen water was used) (Rotavapor-R, Büchi, Switzerland) followedby lyophilization (FreeZone 4.5 L Benchtop Freeze Dry System,Labconco, Kansas, MO, USA). The dry particulate matter (DPM),which refers to the residue remaining after the evaporation of theextraction solvent, was reconstituted in DMSO to give a final con-centration of 15 mg/mL of DPM for the water, MeOH and EA extractsand 5 mg/mL for the DCM extract. A more dilute solution of the DCMextract was necessary to maintain a manageable volume since verylittle DPM was obtained. For each different solvent a new set ofCambridge filters with collected material from CGA burning wasutilized. An “extract” resulted from pooling the extraction solutionsof three Cambridge filters unless otherwise stated. Extracts werethen aliquoted into 1.5 mL vials and stored in the dark at −80 ◦C. Alltoxicity and chromatography experiments using the extracts wereperformed in duplicate, unless stated otherwise.

2.4. Mammalian cell cultures

The cellular lineage used for the IVMNT assay was an interna-tionally registered V79 Chinese hamster cell line (lung fibroblast)obtained from the European Collection of Cell Cultures (V7986041102 lot 04/C/016). Cells were cultured in complete culturemedium (Dulbecco Modified Eagle Medium, DMEM; Gibco, GrandIsland, NY, USA) supplemented with 10% (v/v) heat-inactivatedfetal bovine serum (FBS) and 0.5% (v/v) penicillin/streptomycin(50 units/mL, 50 �g/mL), both from Gibco. Cells were re-suspendedby trypsinization (0.1% trypsin, 1.06 mM EDTA; Gibco) at 37 ◦C.Subcultivation of cells was performed two times per week((1.0–2.0) × 105 cells) into a 75 cm3 Corning flask.

2.5. In vitro micronucleus test (IVMNT)

The IVMNT was performed with V79 Chinese hamster fibroblastcells without metabolic activation (S9 fraction). Cells were grown

in 25 cm3 flasks at a concentration of 5.0 × 105 cells/mL in 10 mLof DMEM for 24 h. The culture medium was then replaced by theDMSO-dissolved extracts added to DMEM at the following con-centrations to which the cells were exposed for 3 h: 5, 10, 15 and20 �g of DPM (or TPM) per mL of DMEM. The positive control was

ogr. A

mcwaldcsL((sTwtd

wtnbC

C

w

%erol

]

Tcc

2

HUadia1

cfTbfc2eemQspw1ts

N. Kaur et al. / J. Chromat

itomycin C (MMC, 0.8 �g/mL; Sigma–Aldrich) and the negativeontrol was DMSO (1%, v/v, in DMEM). After the 3 h exposure, cellsere rinsed twice with Hanks’ Balanced Salt Solution (HBSS, Gibco)

nd re-incubated for 17 h in DMEM containing 3 �g/mL cytocha-asin B (which blocks cellular division, but does not block nuclearivision). Cells were harvested by trypsinization, re-suspended inulture medium at 1.0 × 105 cells/mL and centrifuged onto micro-copic slides at 1200 rpm for 8 min using a Cytospin 3 (Shandon,ondon, UK). Slides were then air dried, fixed in 90% methanol9 min at −20 ◦C) and stained with Acridine Orange solution for 30 s12.5 mg/100 mL of 1×-PBS; Sigma–Aldrich). Finally, slides werecored at 400× magnification according to Fenech’s criteria [27].he percentage of micronuclei, which is a measure of genotoxicity,as determined by first selecting 1000 binucleated (BN) cells and

hen counting the number of these having at least one micronucleusetected, as follows:

%Micronuclei

=(

No. of BN cells with one or more micronucle iTotal no. of BN cells

)× 100

here a micronucleus is defined as a particle surrounded by dis-inct borders, having a maximum of one third the size of the mainucleus and lying inside the cytoplasm [28]. The percentage of inhi-ition of cell proliferation was calculated by first determining theytokinesis-block Proliferation Index (CBPI) [27] as follows:

BPI =(

No. BN cells + 2[No. of multi-nucleated cells]Total no. of cells − mitotic cells

)

here multi-nucleated cells are those having three or more nuclei

Inhibition of cell proliferation = 100 −([

mean CBPI sample dosmean CBPI solvent cont

he average and relative standard deviation (RSD) for the per-entages of micronuclei and inhibition of cell proliferation werealculated from duplicate experiments.

.6. Reproducibility study of the CGA combustion

The precision of the combustion of CGA was evaluated byPLC using a Waters 2695 Separation Module with a Waters 715ltra Wisp automatic injector (Milford, MA, USA). Detection waschieved with a Waters 2475 Multi wavelength fluorescence (FL)etector. The instrument was controlled by ChemStation Plus Fam-

ly software version A.08.03 (Agilent Technologies). Separation waschieved on a Spherisorb ODS2 analytical column (5 �m particles,50 mm × 4.6 mm) from Waters.

Reproducibility of the combustion method was determined byomparing the quantity of selected phenolic compounds obtainedrom four different combustions [24,29], but using only 25% of thePM from each. A quarter of each Cambridge filter (one per com-ustion) was extracted with 10 mL of 1% (v/v) aqueous acetic acidor 30 min on an orbital shaker. The four extracts from the fourombustions were each filtered through a 0.45 �m filter, of which,mL was collected for analysis by HPLC/FL. The volume of eachxtract injected was 10 �L. Separation was achieved by gradientlution (0–100% ACN in 1%, v/v, aqueous acetic acid over 46 min) at aobile phase flow rate of 1.2 mL/min. The total run time was 66 min.uantification was achieved by external calibration as follows. A

tock solution of 1.00 mg/mL of each standard compound was pre-

ared in 1% (v/v) aqueous acetic acid. From the stock solutions, sixorking solutions, ranging from 0 to 50 �g/mL, were prepared in% (v/v) aqueous acetic acid, filtered through a 0.45 �m filter andransferred into 2 mL amber vials. A 20-�L volume of each workingolution was injected in duplicate and a standard calibration curve

1216 (2009) 4703–4712 4705

× 100)

was made by plotting the concentration of the working solutionsversus their respective peak areas.

2.7. Reproducibility study of the extraction with DMSO and DCM

HPLC/MS was used to assess the precision of the DMSO andDCM extraction procedure. The instrument consisted of an 1100HPLC system (Agilent Technologies, Waldbronn, Germany) directlyinterfaced with an Agilent LC/MSD electrospray ionization singlequadrupole mass spectrometer. Injections of 5 �L (75 �g of prod-uct per injection) of DMSO or DCM extracts (the latter havingbeen re-suspended in DMSO) were made onto an Eclipse XDB-C18 analytical column (5 �m particles, 150 mm × 4.6 mm) fromAgilent Technologies. Separation was achieved using a gradientelution of 0–80% MeOH in 0.1% (v/v) aqueous formic acid over24 min at a flow rate of 0.5 mL/min. The total run time was 30 min.For mass spectrometric detection, ions were generated in nega-tive electrospray mode with 4000 V applied on the capillary. Thefragmentor was set at 70 V and the drying gas (N2) was heatedat 300 ◦C and run at 10 L/min. Spectra were acquired from m/z 75to 575 at a rate of 0.94 s/cycle. The reproducibility of the methodof extraction by DMSO and DCM was determined by compar-ing the peak areas (for duplicate injections) of the following 13phenolic reference compounds consistently found in the four differ-ent DMSO extracts: catechol, hydroxyquinone, 4-methyl catechol,4-vinyl catechol, 2-hydroxybenzoic acid, 4-hydroxybenzoic acid, 4-ethyl catechol, 1-(3,4-dihydroxyphenyl) ethanone, p-coumaric acid,coumaric acid, hydrocaffeic acid, quinic acid and caffeic acid methylester.

2.8. Analytical separation of the DMSO extract

Accurate mass-based identification of several products found inthe DMSO extract was achieved using an Agilent 1100 HPLC sys-tem directly interfaced with a 6120 series electrospray ionizationtime-of-flight (TOF) mass spectrometer from Agilent Technologies.The LC/TOF-MS instrument was controlled by Agilent Mass Huntersoftware, and the data was processed by Analyst QS software (Agi-lent Technologies/Sciex). Samples were diluted 1:100 in HPLC gradewater and 2 �L aliquots (0.3 �g of product per injection) wereinjected onto the Eclipse XDB-C18 analytical column. The chro-matographic separation was performed in gradient mode (0–80%MeOH in 0.1%, v/v, aqueous formic acid over 45 min) at a flow rateof 0.5 mL/min. The total run time was 60 min. For MS detection,ions were generated in negative electrospray mode with 4000 Vapplied on the capillary. The fragmentor was set at 200 V and theheated drying gas (N2 at 350 ◦C) was run at 12 L/min. Spectra wereacquired from m/z 50 to 1000 at a rate of 0.94 s/cycle.

2.9. Preparative fractionation of the DMSO extract

The LC system used for preparative fractionation of the DMSOextract consisted of a Gilson 215 LC Handler with 156 UV-Visabsorbance detector (Middletown, WI, USA) directly interfacedwith an LCQ single quadrupole mass spectrometer from ThermoFisher Scientific (Waltham, MA, USA). The instrument was con-

trolled by XCalibur software, version 1.3 (Thermo Fisher) andGilson Unipoint software. DMSO extracts (15 mg/mL) were injected(1.8 mL) and separations were performed on a Prevail C18 prepar-ative column (5 �m particles, 250 mm × 22 mm) from Alltech(Lexington, KY, USA) by gradient elution (0–80% MeOH in 0.1%,

4 togr. A

vT2etg01sflreir5am3fA(cc

2

tmbttlamme1G(fci7wawpioa

2

adpaPf(tpot

706 N. Kaur et al. / J. Chroma

/v, aqueous formic acid over 20 min) at a flow rate of 15 mL/min.he total run time was 30 min and the UV signal was recorded at54 nm concomitant to monitoring the MS signal. Fractions of 8 mLach were collected every 39.1 s into borosilicate disposable cul-ure tubes (10 mm × 100 mm; Fisher Scientific) and then pooled toive three major fractions spanning the following time intervals:–14.2 min, 14.2–23.5 min and 23.5–30 min. A second injection of.8 mL (27 mg) was treated identically and pooled with the corre-ponding major fractions from the first injection in round bottomasks. The three (pooled) fractions were reduced in volume using aotary evaporator for approximately 10 min at 30 ◦C under a mod-rate rotation speed. The flasks were then immersed and rotatedn acetone/dry-ice to induce uniform sample freezing. Finally, theemaining liquids were lyophilized overnight and re-suspended in0% MeOH (aq.), transferred into pre-weighed vials which weregain rotavapped, lyophilized and weighed to obtain the correctass for each fraction. The quantities of the products obtained were

9.5, 22.5 and 7.9 mg, respectively, for the first through third pooledractions. The fractions were stored at −80 ◦C in clear glass vials.pproximately 29% more material was collected than was injected

69.9 mg collected versus 54 mg injected, by calculation). This dis-repancy is probably due to residual DMSO in the first fraction thatannot be entirely evaporated by lyophilization.

.10. Chemical characterization and separation of fraction 2

The LC/TOF-MS described above, which is a high resolu-ion system, was used for the chemical characterization of the

ost bioactive fraction of the DMSO extract. This was achievedy first using the lower resolution LC/MSD system (see sec-ion in Reproducibility studies) to optimize the separation of aest mixture representative of fraction 2, comprised of the fol-owing seven standards: caffeic acid, benzoic acid, p-coumariccid, p-hydroxybenzoic acid, trans-cinnamic acid, 4-hydroxy-3-ethoxycinnamic acid and 2,5-dihydroxybenzoic acid. This testixture was injected onto four different stationary phases: Syn-

rgi Polar-RP (4 �m, 150 mm × 4.6 mm), Synergi Hydro-RP (4 �m,50 mm × 4.6 mm), Gemini C18 (5 �m, 150 mm × 4.6 mm) andemini C6-Phenyl (5 �m, 150 mm × 4.6 mm), all from Phenomenex

Torrance, CA, USA). Separations were carried out under nine dif-erent gradient elution conditions by varying the initial MeOHoncentration as follows: 10, 15, 20, 25, 30, 35, 40, 45 and 50%,n 0.1% (v/v) aqueous formic acid, with the gradient applied up to5%, over the first 24 min in each case. The best gradient conditionsere transferred to the higher resolution LC/TOF-MS instrument

nd applied to the separation of fraction 2 components. Samplesere first diluted 100-fold in 50% MeOH (aq.) to make them com-atible with the dynamic range of the LC/TOF-MS system, then

njections of 2 �L (corresponding to 0.3 �g of product) were maden the four columns listed above. The total run time was 30 min atflow rate of 0.5 mL/min.

.11. Preparative sub-fractionation of fraction 2

To sub-fractionate “fraction 2” of the DMSO extract by prepar-tive LC, an injection of 2.0 mL was made on the instrumentescribed for preparative fractionation of the DMSO extract. Sam-les [4.24 mg/mL in 75%, v/v, MeOH (aq.)] were injected in duplicatend separations were performed on an AXIA packed Synergiolar-RP preparative column (4 �m particles, 100 mm × 21.2 mm)rom Phenomenex. A Polar-RP security guard prep cartridge

15 mm × 21.2 mm) from Phenomenex was installed upstream ofhe preparative column. The chromatographic separation waserformed in gradient mode (15–75% MeOH in 0.1%, v/v, aque-us formic acid over 20 min) at a flow rate of 6 mL/min. Theotal run time was 30 min and the UV signal was monitored at

1216 (2009) 4703–4712

254 nm concomitant with the MS signal. Fractions of 4 mL eachwere collected every 19.8 s into borosilicate disposable culturetubes (10 mm × 100 mm; Fisher Scientific) and then pooled to givefour large sub-fractions spanning the following time intervals:12.0–15.2, 15.2–16.4, 16.4–21.2 and 21.2–30 min. Each pooled sub-fraction was placed in a round-bottom flask and was treated asdescribed above during the first fractionation step. The amounts ofproduct obtained for the first through fourth pooled sub-fractionswere 2.16, 1.49, 5.87 and 9.07 mg respectively. A small portion of(major) fraction 2 of the DMSO extract was used for control studies.Samples were kept at −80 ◦C in clear glass vials until utilization.

2.12. Chemical characterization of sub-fraction 1

The LC/TOF-MS system described above was used for the accu-rate mass identification of products present in the sub-fractiondisplaying the highest toxicity. Samples were diluted 1:100 in 50%(v/v) MeOH (aq.) and injections of 2 �L aliquots were performed onthe Polar-RP column (4 �m, 150 mm × 4.6 mm) followed by separa-tion by gradient elution (0–80% MeOH in 0.1%, v/v, aqueous formicacid over 24 min) at a flow rate of 0.5 mL/min. The total run timewas 15 min.

2.13. Statistical analysis

The results for the combustion reproducibility study weretested for comparison of linearity between different groups ofeither extracts or fractions using the analysis of covariance(ANCOVA) method. For comparison between the DMSO extracts,the percentages of micronuclei (genotoxicity) and inhibition ofcell proliferation were taken as the direct quantitative variable,the dose of exposure as the quantitative dependent covariableand the extraction solvent was taken as the qualitative covari-able for two replicates. ANCOVA compares the dose–responselinearity between each extract. Significant differences betweenextracts were determined by the Duncan’s multiple comparisontest and were considered significant when p < 0.05. Toxicologicaldata obtained from the IVMNT for the different solvent extracts andfractionation studies were analyzed using XLSTAT software, version7.5 (Addinsoft Brooklyn, NY, USA).

Analysis of variance (ANOVA) was used to evaluate the toxicityresults where the dose, the CGA extracts, the fractions and the sub-fractions were all considered as factors. The dose by extract/fractioninteraction was also included in the model. In order to assess dif-ferences between the CGA extracts/fractions for the different doses,the dose by extract interaction was investigated using multiplecomparisons. More specifically, the extracts/fractions were ana-lyzed by the Fisher least significant difference multiple comparisontest with a Bonferroni correction to type 1 error to ensure that theoverall risk was kept under ˛ = 5%. In all cases, the background levelof genotoxicity generated by the control solvent (1% DMSO) wassubtracted from the micronuclei percentage values for all samples.As a result of the statistical analysis, the data were grouped as fol-lows: A, B or C. Samples sharing the same letter, i.e., lie within thesame group, are not statistically different.

3. Results and discussion

The various toxicological studies carried out on tobacco smokehave been generally related to the combustion products of wholetobacco [30–32]. The aim of our study was to characterize the

toxicity of the combustion products of one individual tobaccocomponent, CGA, which is the major polyphenolic component oftobacco. A few toxicological studies have reported on the genotoxi-city of CGA [14] and its incomplete combustion products [13,19,21].In addition, some chemical studies have been published on the

N. Kaur et al. / J. Chromatogr. A 1216 (2009) 4703–4712 4707

Table 1Reproducibility of the partial combustion process of CGA: phenolic compounds identified by HPLC/FL in 1% (v/v) aqueous acetic acid extracts of the TPM from 1/4 of each offour Cambridge filters.

CGA combustion replicates TPM qty on 1/4 Cambridge filter (mg) Phenolic content (�g/mg TPM) Average

Hydroquinone Resorcinol Catechol Phenol p-Cresol

�ex (nm) – 285 270 270 270 270�em (nm) – 325 310 310 298 305Combustion 1 27.8 16.1 0.8 32.9 10.2 0.1Combustion 2 33.2 19.5 1.0 40.4 16.1 0.1Combustion 3 32.9 19.2 1.0 34.1 10.4 <LODCombustion 4 38.2 15.1 0.9 31.2 9.1 0.1A 5R 6

ipscrwooJo

3

ieaebtat

coobmtotsTpsatc

TN

C

CCCCASR

verage 33.0 17.SD (%) 12.9 12.

a N/C: not calculated.

dentification of CGA combustion products [10,12]. However, norevious study on relating genotoxicity to the chemical compo-ition of the combustion products of CGA has been made. Theombustion conditions used in this study were chosen based on theange of temperatures found during the combustion of cigarettes,hich occur between 300 and 900 ◦C and higher [33]. The precision

f our chemical analyses required a robust and reproducible meansf simulating the partial combustion of CGA, which was why thePTP apparatus was employed. Furthermore, this study was carriedut at a single combustion temperature of 640 ◦C for simplicity.

.1. Reproducibility study of the CGA partial combustion process

In order to understand and quantify any variability in the tox-cological and/or chemical analyses, it was deemed important tovaluate the precision of the CGA combustion method. This wasssessed by comparing: (a) the phenolic content in four differentxtracts by HPLC/FL and (b) the genotoxicity and the degree of inhi-ition of cell proliferation between three of the four extracts usinghe IVMNT. The Cambridge filters were extracted with 1% (v/v)queous acetic acid for this study because this solution is knowno extract phenolic compounds well [29].

HPLC/FL showed that the concentration of hydroquinone, resor-inol, catechol and phenol (reported as a function of the quantityf TPM extracted per quarter filter) varied with an average RSDf 15.5% (and median RSD of 12.6%) (Table 1). p-Cresol was oftenelow the limit of quantification. The high polarity of the solventay have impeded the extraction of p-cresol, thus explaining why

he latter was barely detected. To evaluate the relative proportionf each compound, their concentrations were normalized relativeo hydroquinone (Table 2) for each experiment to eliminate theampling error associated with extracting only 1/4 of the filter pad.he relative (i.e., normalized) concentrations of the phenolic com-

ounds resorcinol, catechol and phenol determined by HPLC/FL,howed an average of 11.1% RSD (Table 2). This precision is in closegreement with the HPLC/FL determination of phenols in the par-iculate phase of mainstream cigarette smoke of the 1R5F referenceigarette reported recently by Moldoveanu and Kiser [34]. As seen

able 2ormalized phenolic content relative to hydroquinone (from Table 1).

GA combustion replicates Normalized quantity relative to hydroquinone

Hydroquinone Resorcinol

ombustion 1 1.00 0.052ombustion 2 1.00 0.053ombustion 3 1.00 0.053ombustion 4 1.00 0.060verage – 0.054D – 0.004SD (%) – 7.1

a N/C: not calculated.

0.9 34.7 11.5 0.110.4 11.6 27.5 N/Ca 15.5

in Table 1, the concentration of phenol varied the most amongthe four combustions; its RSD was more than twice that of theother phenolic compounds. Although phenol is the most volatileof the five species, ineffective trapping was ruled out as a sourceof its high variability because the temperature did not exceed 45 ◦Cat the Cambridge filter pad position. Although the experimentalprocedure was identical for each sample, it is possible that slight dif-ferences in the rotavaporation step may have contributed to lossesof phenol in some samples. Calibration using an internal standardsuch as 4-chlorophenol deposited on the Cambridge filter pad priorto extraction [34] would be necessary to confirm this.

The IVMNT method was chosen to measure the extracted TPMbioactivity because it is one of the in vitro toxicity tests recom-mended for tobacco smoke studies by the Cooperation Centrefor Scientific Research Relative to Tobacco (CORESTA) and HealthCanada [23,25,26]. Overall, although a variation of 11.1% in nor-malized phenolic content was present between the extracts, thisdid not translate into a similar variation in bioactivity. The IVMNTdata (Fig. 1a and b) showed that the percentages of micronucleiand inhibition of cell proliferation among different extracts werenot significantly different as per the Duncan’s multiple compari-son test. Therefore, we decided to continue with this method ofcombustion using the JPTP. However, to reduce the impact of thehigh variability between combustions, we pooled the extracts fromthree independent combustions to obtain one final pooled extract,which was then divided into equal aliquots and stored at −80 ◦C forsubsequent toxicological and chemical assays.

3.2. Effect of extraction solvent

Selective solvent extraction was used to initiate the chemi-cal characterization study of CGA combustion products. The fivesolvents, used in parallel, were DMSO [polarity index (P) = 7.2,

dipole moment (DM) = 3.96], water (P = 9, DM = 1.85), MeOH (P = 5.1,DM = 1.70), DCM (P = 3.1, DM = 1.60) and EA (P = 4.4, DM = 1.78), thusyielding five different extracts. These solvents were chosen dueto their different polarity index values and because of the lim-ited selectivity and high variability observed with 1% (v/v) aqueous

Average

Catechol Phenol p-Cresol

2.042 0.633 0.0062.076 0.826 0.0071.775 0.539 0.0012.067 0.604 0.0051.990 0.651 0.0050.144 0.124 0.0037.2 19.0 N/Ca 11.1

4708 N. Kaur et al. / J. Chromatogr. A 1216 (2009) 4703–4712

Fig. 1. The genotoxic activity (% micronuclei) (a), and inhibition of cell proliferation(b), induced by DMSO extracts from three independent combustions of 0.5 g of CGA,on V79 cells exposed for 3 h without metabolic activation. Dose refers to the quan-tDCd

aspiibfpifipc

owiaeDecagoteTrha

eb

Fig. 2. The genotoxic activity (% micronuclei), (a), and inhibition of cell proliferation,(b), induced by DCM, DMSO, water, EA and MeOH extracts generated from combus-tions of 0.5 g of CGA, on V79 cells exposed for 3 h without metabolic activation. Doserefers to the quantity of TPM (�g) per mL of medium. n = 2 for all the condensatesexcept DCM and DMSO, where n = 4. The dose/extraction solvent interactions were

Four independent combustion experiments were carried out forboth DMSO and DCM. Each extract obtained was injected in dupli-cate. The abundance (peak areas) of 13 reference compounds foundin the extracts were monitored (Table 3). Retention times were

Table 3Reproducibility of the extraction by DMSO (n = 4), with respect to peak area for 13phenolic compounds identified by LC/MSD (negative mode).

Reference product name Average peak area (103) RSD (%)

Catechol 19 ± 1 5.3Hydroxyquinone 400 ± 30 7.54-Methyl catechol 29 ± 2 6.94-Vinyl catechol 710 ± 40 5.62-Hydroxybenzoic acid 160 ± 10 6.34-Hydroxybenzoic acid 310 ± 20 6.54-Ethyl catechol 210 ± 20 9.51-(3,4-Dihydroxyphenyl)ethanone 25 ± 2 8.0

ity of TPM or DPM (�g) per mL of medium. The three extracts were tested in theuncan’s test (˛ = 0.05) and no statistically significant difference was found. (· · ·�· · ·)ombustion 1; (- -�- -) combustion 2; (–�–) combustion 3. Error bars indicate stan-ard deviation.

cetic acid as an extraction solvent. Schlotzhauer and Chortykhowed that tobacco directly extracted with solvents of variousolarity or “extraction strength” yielded extracts of different chem-

cal composition [35]. According to their miscibility and polarityndex, the most hydrophilic products are preferentially extractedy DMSO, water and MeOH, whereas less hydrophilic products areound in the DCM and EA extracts. Generally, the phenolic com-ounds have amphiphilic properties, and thus should be found

n every extract. In a previous work [36], GC/MS analysis of theve extracts of CGA combustion products showed the presence ofhenolic compounds such as catechol, phenol, hydroquinone, ethylatechol, benzoic acid and quinic acid in most of the extracts.

Based on their chemical composition and the relative amountsf each combustion product, certain extracts among the five testedere expected to induce a higher degree of genotoxicity and/or

nhibit cell proliferation than others by the IVMNT. The percent-ge of micronuclei, or genotoxicity, is shown in Fig. 2a for the fivextracts. At a dose of 20 �g/mL, the genotoxicity induced by theCM extract was significantly different from the water and MeOHxtracts, but not from the DMSO and EA extracts. The inhibition ofell proliferation is shown in Fig. 2b for the five extracts. At doses of 5nd 10 �g/mL there was no statistically significant difference in theenotoxicity induced, as evaluated by ANOVA. However, at dosesf 15 and 20 �g/mL, the inhibition of cell proliferation induced byhe DCM extract was significantly different from all other extractsxcept DMSO at 15 �g/mL, and except water and DMSO at 20 �g/mL.he negative values observed for the inhibition of cell proliferationeflect cell growth. Overall, the DMSO and DCM extracts inducedigher biotoxicity compared to extracts obtained using water, EA

nd MeOH.

DCM is the least polar of the solvents tested and thus would bexpected to extract phenolic compounds, which are known to beioactive. DMSO on the other hand possesses excellent solvating

analyzed by the Fisher least significant difference multiple comparison test with aBonferroni correction to type 1 error to ensure that the overall risk was kept under˛ = 0.05. Letters A, B and AB designate different statistical groups. Error bars havebeen removed for clarity.

powers; it dissolves both polar and non-polar compounds. Further-more, a low concentration of DMSO (1%, v/v, in DMEM) has lowtoxicity [37], which was why the other extraction solvents werereconstituted in DMSO for the IVMNT assays.

3.3. Reproducibility study of the extraction with DMSO and DCM

Based on the results comparing extraction by five different sol-vents, the precision of the DMSO and DCM extraction procedureswas evaluated by LC/MS to ensure a robust and reliable method.

p-Coumaric acid (isomer 1) 170 ± 10 5.9Coumaric acid (isomer 2) 34 ± 3 8.8Hydrocaffeic acid 120 ± 6 5.0Quinic acid 54 ± 3 5.6Caffeic acid methyl ester 30 ± 3 10.0

N. Kaur et al. / J. Chromatogr. A 1216 (2009) 4703–4712 4709

F produb cted.

hfenelnaseapg

3

tDacgf

wocpgtowtitltch

ig. 3. Total ion chromatogram of the DMSO extract showing the CGA combustioneen overlaid to represent the compounds isolated in the three main fractions colle

ighly reproducible (≤0.1% RSD) across the four extracts testedor both extraction solvents. The peak area precision of the DMSOxtraction (<10% RSD) was nine times better than that of DCM (dataot shown). This may have been due to the volatile nature of DCM;vaporation may have occurred during the extraction procedureeading to less reproducible results. Therefore, DCM extraction wasot further investigated. In addition, DMSO was observed to extractlarger number of compounds, which is in keeping with its good

olvating strength. With respect to biological activity, the DMSOxtracts were not further tested by the IVMNT since the resultsbove showed that variation in genotoxicity and inhibition of cellroliferation was minimal even though phenolic content variedreatly (15.5% average RSD, Table 1).

.4. Analytical separation of the DMSO extract

Accurate mass determination by LC/TOF-MS was used to iden-ify the main components, and class of components, in the wholeMSO extract (Fig. 3). Over 40 compounds were identified by neg-tive ionization mode, which was used because the majority of theombustion products possessed alcohol and/or acidic functionalroups. These results guided the choice of which fractions to poolor preparative LC.

Several phenolic compounds were present in the DMSO extract,hich is consistent with previous studies of CGA [10,12,13]. Based

n the complexity of the combustion products of a single tobaccoomponent like CGA, we can only begin to imagine the com-lexity of whole tobacco smoke. Only techniques like LC/MS andas chromatography–MS [34,38] have the selectivity and resolu-ion needed to provide reliable identification and quantificationf such a large range of components. Although the DMSO extractas bioactive according to the IVMNT, it was difficult to iden-

ify the specific compounds responsible for bioactivity. Therefore,t was necessary to further simplify the extract. Some poten-

ial techniques to achieve this include: filtration, centrifugation,iquid–liquid extraction, solid-phase extraction and sample frac-ionation, among others. Fractionation by preparative scale LC washosen based on its ability to divide the sample into precise portionsaving sufficient quantity for further analysis by the IVMNT.

cts, which were analyzed by LC/TOF-MS in negative mode. The dashed lines have

3.5. Preparative fractionation of the DMSO extract

The DMSO extract was fractionated by preparative LC/UV (detec-tion at 254 nm) into three major fractions, as indicated by thedotted lines in Fig. 3. This allowed for determination of the differ-ence in toxicity between fractions and presumably a convergenceon the compounds responsible for the observed toxicity. The firstfraction, which was selected to include quinic acid-related com-pounds and other non UV-absorbing species, contained 39.5 mg ofproduct. The second fraction (22.5 mg) included catechol and itsderivatives while the third (7.9 mg) comprised more hydrophobiccompounds. Biotoxicity was assessed by the IVMNT. As illustrated(Fig. 4a and b), among the three major fractions tested, fraction2 induced the highest percentages of micronuclei and inhibitionof cell proliferation compared to fractions 1 and 3. The increasedlevel of toxicity generated by fraction 2 was likely due to thepresence of phenolic compounds found in that fraction. Fig. 4ashows that the whole DMSO extract as well as the second fractioninduced the highest percentage of micronuclei. However, only the(whole) DMSO extract induced a significantly higher percentage ofmicronuclei at a dose of 35 �g/mL. Fig. 4b shows that fraction 2and the (whole) DMSO extract induced a significantly higher inhi-bition of cell proliferation compared to fractions 1 and 3 at doses of15–35 �g/mL.

3.6. Chemical characterization and separation of fraction 2

Due to its overall higher bioactivity, the second fraction was re-analyzed by LC/TOF-MS with accurate mass measurement to assessits chemical composition. Fraction 2 was found to contain catecholand its derivatives (methyl catechol, ethyl catechol and vinyl cate-chol), phenol, hydrocaffeic acid, 1-(3,4-dihydroxyphenol)ethanone,3,4-dihydroxybenzoic acid, p-hydroxycinnamic acid, p-coumaricacid, caffeic acid methyl ester, caffeic acid and hydroxybenzoic acid

(Fig. 3). Among these, the last 8 compounds (hydrocaffeic acid tohydroxybenzoic acid) have not been previously reported as carcino-gens, mutagens or teratogens as opposed to catechol, phenol andcaffeic acid [39]. As previously discussed, catechol and its deriva-tives are known to be responsible for induction of micronuclei and

4710 N. Kaur et al. / J. Chromatogr. A 1216 (2009) 4703–4712

F(E

tf

ribmousadTts

2 of the DMSO extract were transferred to a Polar-RP preparative

Ft

ig. 4. The genotoxic activity (% micronuclei), (a), and inhibition of cell proliferationb), induced by (whole) DMSO extract, fractions 1–3. All other conditions as in Fig. 2.rror bars have been removed for clarity.

oxicity in the micronuclei assay [18] and thus could be responsibleor the increased level of bioactivity of fraction 2.

The analytical separation of fraction 2 was optimized withespect to peak resolution with the objective of sub-fractionatingt for further analysis to identify the compounds responsible for itsioactivity. Based on the compounds identified in fraction 2, a testixture of seven standard compounds was prepared and a series

f different stationary phases and eluent compositions were eval-ated on the LC/MSD instrument as described in the Experimentalection. The best resolution for the test mixture was obtained with15 (or 20) to 75% MeOH in 0.1% (v/v) aqueous formic acid gra-

ient over 23 min using the Polar-RP column (data not shown).his column, which is composed of an ether-linked phenyl sta-ionary phase with polar end-capping, most likely enabled a moreelective interaction with the aromatic compounds and improved

ig. 5. Preparative LC chromatogram (254 nm UV trace) of fraction 2 of the DMSO extract she four main sub-fractions collected.

Fig. 6. The genotoxic activity (% micronuclei) (a), and inhibition of cell proliferation(b), induced by DMSO extract, fraction 2 and sub-fractions 1–4. All other conditionsas in Fig. 2. Error bars have been removed for clarity.

their resolution. Subsequently, fraction 2 was analyzed under theoptimized conditions by LC/TOF-MS. This enabled separation ofthe quite abundant and bioactive catechol from three isomers ofhydroxybenzoic acid (data not shown).

3.7. Preparative sub-fractionation of fraction 2 and chemicalcharacterization of sub-fraction 1

The optimized analytical separation conditions used for fraction

column for sub-fractionation. Fig. 5 shows how we generated thefour major sub-fractions of fraction 2 by preparative LC. The geno-toxicity induced by these four sub-fractions, as well as by majorfraction 2 and by the whole DMSO extract is shown in Fig. 6a.

howing the sub-fractions collected. The dotted lines have been overlaid to represent

N. Kaur et al. / J. Chromatogr. A 1216 (2009) 4703–4712 4711

F hromatograms (lower traces) of sub-fraction 1 of fraction 2 of the DMSO extract of CGAc

NA(1tDiDet2ti

adwlcipiodTfc

nemfiF

ig. 7. Base peak LC/TOF-MS chromatogram (upper most trace) and extracted ion combustion products. Separation conditions are given in Section 2.

o statistically significant difference (˛ = 0.05) was measured byNOVA between the six samples compared at the lower dose range

5 and 10 �g/mL). Whereas, at the dose ranges corresponding to5 and 20 �g/mL there were statistically significant differences inerms of generation of micronuclei between sub-fraction 4, theMSO extract and sub-fraction 2. The inhibition of cell proliferation

nduced by the four sub-fractions, by fraction 2 and by the wholeMSO extract is compared in Fig. 6b. Sub-fraction 1 and the DMSOxtract induced a higher percentage of inhibition of cell prolifera-ion but were only significantly higher compared to sub-fractionsand 3 at doses of 15 and 20 �g/mL. Overall, the IVMNT showed

hat sub-fraction 1 induced the highest degree of genotoxicity andnhibition of cell proliferation compared to the other sub-fractions.

Subsequently, sub-fraction 1 was analyzed by LC/TOF-MS (Fig. 7)nd found to contain catechol as the major component, 3,4-ihydroxybenzoic acid and a third, less abundant compoundith the empirical formula C6H8O2. Based on this formula, some

ogical structures were deduced. One possible compound is 1,2-yclohexanedione, for which no toxicology information was foundn the literature A set of standards of 1,2-cyclohexanedione wererepared, but they were inactive in terms of toxicological response

n the dose range of 5–20 �g/mL. A second possibility may be onef the isomers of dihydroxycyclohexadiene. Unfortunately, no stan-ards were commercially available to test biotoxicity by the IVMNT.o the best of our knowledge, toxicological data is also not availableor any of these isomers. Further structural analysis of the C6H8O2ompound was beyond the scope of this study.

The second compound identified, 3,4-dihydroxybenzoic acid, is

ot known to be either genotoxic or an inhibitor of cell prolif-ration [39]. Catechol, on the other hand, which was ca. 10-foldore abundant than 3,4-dihydroxybenzoic acid (Fig. 7), was con-

rmed to be genotoxic and inhibit cell proliferation as seen inig. 8a and b for catechol standards (5–20 �g/mL dose range)

Fig. 8. The genotoxic activity (% micronuclei) (a), and inhibition of cell proliferation(b), induced by DMSO extract and catechol standards. All other conditions as in Fig. 1.(–�.–) DMSO; (- -�- -) catechol standard. Error bars indicate standard deviation.

4 togr. A

aieVtttocpotgssc

4

tstblaiumDawawbtbaoetiat

mtpmtcttrttmt

[

[

[

[

[[[[[

[[[[[[[[[

[

712 N. Kaur et al. / J. Chroma

ssessed by the IVMNT. These results support previous findingsn terms of the toxicological response [40] and in terms of cat-chol being a product of the combustion of CGA [11,13,16,18,19].aughan et al. recently reported that mainstream smoke from

he 1R4F reference cigarette contained essentially the same quan-ity of catechol as hydroquinone per cigarette, whereas flue-curedobacco cigarettes delivered an amount of catechol twice thatf hydroquinone [38]. We also obtained a 2-to-1 ratio of cate-hol to hydroquinone in the extracted particulate matter afterartial combustion of CGA (Table 2). This suggests that CGA isne of the major sources of phenolic components in flue-curedobacco [38], thus demonstrating that our choice to investi-ate the burning of the single tobacco component CGA is aimpler yet valid alternative to using whole tobacco smoke totudy the relationship between toxicity and fractional chemicalomposition.

. Conclusion

A multidisciplinary study comprising solvent extraction, frac-ionation, bioassay and state-of-the-art LC/MS allowed us toystematically narrow in on the biotoxic components in the par-iculate matter produced from the incomplete combustion (i.e.,urning) of chlorogenic acid (CGA). Extraction with DMSO fol-

owed by successive chromatographic fractionation combined withccurate mass identification and use of the IVMNT for bioactiv-ty identified catechol, 3,4-dihydroxybenzoic acid and a minor,nidentified constituent (C6H8O2) as being components of theost bioactive sub-fraction of CGA combustion products. 3,4-ihydroxybenzoic acid has not been reported to be genotoxic orn inhibitor of cell proliferation. Catechol, on the other hand,as the major component present in the most toxic sub-fraction

nd is known to be toxic. By testing catechol standards alone,e were able to confirm that catechol is indeed genotoxic andlocks cell proliferation in the dose working range. We suspecthat catechol is therefore the major component responsible for theioactivity resulting from the whole DMSO extract. Furthermore,positive correlation was established between CGA (compared tother polyphenolic compounds) found in tobacco and catechol andthyl-catechol found in smoke [11,13]. This demonstrates that inerms of chemistry, our approach of studying a single components not only valid but is also relevant. The relationship between CGAnd catechol would support the reduction of CGA in tobacco in ordero reduce catechol.

Our research carried out on the combustion products of CGAay not be directly correlated to the smoke from all cigarette

ypes due to the fact that the combustion of a single tobacco com-onent does not take into account possible interactions betweenultiple components during combustion. Also, the conditions of

obacco combustion, such as heating rate and atmospheric gasoncentration have been shown to influence the relative propor-ions of the products [33]. However, our methodology allows forhe analysis of a simpler product mixture. Also, we cannot directly

elate the toxicological results from the in vitro assays to in vivooxicity since there are detoxification pathways involved in the lat-er. Finally, only the compounds detected by LC/MS in negative

ode were accounted for. Nonetheless, our approach combiningoxicology with chemical identification has contributed to a bet-

[

[

[

1216 (2009) 4703–4712

ter understanding of the toxicity of a single tobacco component,CGA.

Acknowledgements

Graduate bursaries for Navneet Kaur were provided by theNatural Science and Engineering Council of Canada (NSERC) andCORESTA. We thank N. Poirier for help with culture maintenanceand the IVMNT, and J. Dumont for help with the HPLC/FL analyses.We also thank D. Sekhon and K. Venne for their assistance with thevarious LC/MS instruments at U. of Montréal and M. Bratberg forher contribution to LC/MSD method development.

References

[1] A. Rodgman, T.A. Perfetti, The Chemical Components of Tobacco and TobaccoSmoke, CRC Press, Boca Raton, FL, 2008.

[2] A. Rodgman, C.R. Green, Beitr. Tabakforsch. Int. 20 (2003) 481.[3] D. Hoffmann, I. Hoffmann, J. Toxicol. Environ. Health 50 (1997) 307.[4] T. Jansson, M. Curvall, A. Hedin, C.R. Enzell, Mutat. Res. 169 (1986) 129.[5] T. Paschke, G. Scherer, W.D. Heller, Beitr. Tabakforsch. Int. 20 (2002) 107.[6] R.L. Stedman, Chem. Rev. 68 (1968) 153.[7] G.L. Huber, Semin. Respir. Med. 10 (1989) 297.[8] V.C. Runeckles, Can. J. Biochem. Physiol. 41 (1963) 2249.[9] E.L. Wynder, D. Hoffmann, Tobacco and Tobacco Smoke, Studies in Experimental

Carcinogenesis, Academic Press, New York, 1967.[10] R.K. Sharma, T.S. Fisher, M.R. Hajaligol, J. Anal. Appl. Pyrol. 62 (2002) 281.[11] W.S. Schlotzhauer, R.M. Martin, M.E. Snook, R.E. Williamson, J. Agric. Food.

Chem. 30 (1982) 372.12] H. Sakuma, S. Matsushima, S. Munakata, S. Sugawara, Agric. Biol. Chem. 46

(1982) 1311.[13] W.S. Schlotzhauer, M.E. Snook, O.T. Chortyk, R.L. Wilson, J. Anal. Appl. Pyrol. 22

(1992) 231.[14] H.F. Stich, M.P. Rosin, C.H. Wu, W.D. Powrie, Mutat. Res. 90 (1981) 201.[15] Q. Li, M.T. Aubrey, T. Christian, B.M. Freed, Fund. Appl. Toxicol. 38 (1997) 158.[16] R. Gopalakrishna, Z.H. Chen, U. Gundimeda, Proc. Natl. Acad. Sci. 91 (1994)

12233.[17] J.M. McCue, S. Lazis, J.J. Cohen, J.F. Modiano, B.M. Freed, Mol. Immunol. 39 (2003)

995.[18] M.L. Robertson, D.A. Eastmond, M.T. Smith, Mutat. Res. 249 (1991) 201.[19] D. Hoffmann, S.S. Hecht, E.L. Wynder, Environ. Health Persp. 50 (1983) 247.20] Q. Li, L. Geiselhart, J.N. Mittler, S.P. Mudzinski, D.A. Lawrence, B.M. Freed, Toxicol.

Appl. Pharmacol. 139 (1996) 317.21] D. Préfontaine, A. Morin, C. Jumarie, A. Porter, Food Chem. Toxicol. 44 (2006)

724.22] M. Poirier, M. Fournier, P. Brousseau, A. Morin, J. Toxicol. Environ. Health A 65

(2001) 1437.23] Official Method T-503, Health Canada, Ottawa, 2004.24] Official Method T-114, Health Canada, Ottawa, 1999.25] OECD Publications, OECD, Paris, 2004.26] CORESTA In Vitro Toxicology Task Force, CORESTA, Paris, 2002.27] M. Fenech, W.P. Chang, M. Kirsch-Volders, N. Holland, S. Bonassi, E. Zeiger,

Mutat. Res. 534 (2003) 65.28] W. Frieauff, F. Potter-Locher, A. Cordier, W. Suter, Mutat. Res. 413 (1998) 57.29] C.H. Risner, S.L. Cash, J. Chromatogr. Sci. 28 (1990) 2.30] R.E. Frenesius, J. Anal. Appl. Pyrol. 8 (1985) 561.31] A. Rodgman, Beitr. Tabakforsch. Int. 20 (2003) 402.32] D.M. DeMarini, Mutat. Res. 567 (2004) 447.33] R.R. Baker, Prog. Energy Combust. Sci. 7 (1981) 135.34] S.C. Moldoveanu, M. Kiser, J. Chromatogr. A 1141 (2007) 90.35] W.S. Schlotzhauer, O.T. Chortyk, Tob. Sci. 25 (1981) 6.36] M. Lacasse, Caractérisation toxicologique et chimique des produits de com-

bustion d’un composé pur du tabac: l’acide chlorogénique, Master’s Thesis,University of Québéc/INRS-IAF, Montreal, 2007.

37] R.P. Vignes, Presented at the 220th American Chemical Society National Meet-ing, Washington, DC, August 20–24, 2000.

38] C. Vaughan, S.B. Stanfill, G.M. Polzin, D.L. Ashley, C.H. Watson, Nicotine Tob. Res.10 (2008) 1261.

39] L.S. Gold, E. Zeiger, Handbook of Carcinogenic Potency and GenotoxicityDatabases, CRC Press, Boca Raton, FL, 1997.

40] S. Chouchane, J.B. Wooten, F.J. Tewes, A. Wittig, B.P. Muller, D. Veltel, J. Diek-mann, Chem. Res. Toxicol. 19 (2006) 1602.