Ac

Ka

b

a

ARR2AA

KHTMTPV

1

omaa[abitpag

L

h0

Journal of Chromatography A, 1381 (2015) 229238

Contents lists available at ScienceDirect

Journal of Chromatography A

j o ur na l ho me page: www.elsev ier .com/ locate /chroma

nalysis of triterpenoids and phytosterols in vegetables by thin-layerhromatography coupled to tandem mass spectrometry

aterina Naumoskaa, Irena Vovka,b,

National Institute of Chemistry, Laboratory for Food Chemistry, Hajdrihova 19, 1001 Ljubljana, SloveniaEN-FIST Centre of Excellence, Trg Osvobodilne fronte 13, 1000 Ljubljana, Slovenia

r t i c l e i n f o

rticle history:eceived 14 October 2014eceived in revised form8 December 2014ccepted 1 January 2015vailable online 9 January 2015

eywords:PTLCLCMS2

ass spectral libraryriterpenoidshytosterolsegetables

a b s t r a c t

Three TLC methods were used for an initial screening of some common plant triterpenoids and phytos-terols in cuticular wax extracts of different vegetables (zucchini, eggplant, tomato, red pepper, mangold,spinach, lettuce, white-colored radicchio di Castelfranco, raddichio Leonardo, white cabbage, red cab-bage and savoy cabbage). The preliminary experiments showed that the studied vegetables are potentialsources of triterpenoids and phytosterols. To identify the compounds present in the extracts with high cer-tainty, the first TLCMS2 method was developed for the analysis of eight triterpenoids (lupeol, -amyrin,-amyrin, cycloartenol, cycloartenol acetate, lupeol acetate, lupenone and friedelin) and two phytosterols(-sitosterol and stigmasterol). This method takes the advantages of: (1) a satisfactory separation of thetarget compounds; (2) their differentiation according to the band colors; and (3) the potential of theirdiscrimination by the acquired first-order mass (MS) and product ion (MS2) spectra. Since the closelyeluting compounds have complex and similar MS2 spectra, distinguishing between them was possible bythe proposed characteristic ions. Using a custom-built mass spectral library, the head to tail MS2 spectra

comparison of sample test solution zones and standard aided the compound identification. In addition tothe molecular mass information, the developed atmospheric pressure chemical ionization method (APCI)in positive ion mode provided structural information, regarding the presence of functional group in themolecule. This approach resulted in many positively assigned compounds in the investigated vegetablewaxes, from which more than a half are reported for the first time.

2015 Elsevier B.V. All rights reserved.. Introduction

Cuticular waxes constitute the waxy coverings of the plantrgans and serve as plant protectants against unfavorable environ-ental conditions and insects. They are composed of long-chainliphatic hydrocarbons, ketones, esters, fatty alcohols, fatty acids,ldehydes, as well as triterpenoids (C30) and phytosterols (C18C30)1]. The last two groups of compounds present a very largend structurally diverse family of secondary plant metabolites,iosynthetically derived through the mevalonate pathway from sixsoprene units (C5H8) [2]. They can exist as free compounds or inhe form of esters and saponins. Pharmacological studies of triter-

enoids showed anticancer, anti-inflammatory, anti-ulcerogenic,nti-microbial, anti-viral (including anti-HIV), anti-fungal, anal-esic, antioxidative, hepatoprotective and some other activities

Corresponding author at: National Institute of Chemistry, Hajdrihova 19, SI-1000jubljana, Slovenia. Tel.: +386 1 4760 341; fax: +386 1 4760 300.

E-mail address: irena.vovk@ki.si (I. Vovk).

ttp://dx.doi.org/10.1016/j.chroma.2015.01.001021-9673/ 2015 Elsevier B.V. All rights reserved.[310]. Phytosterols are well known for their cholesterol-loweringproperties, and also demonstrate anticancer, anti-inflammatoryand immunoregulatory activities [11,12]. Due to the growing inter-est in triterpenoids and phytosterols, the development of suitablemodern analytical methods for the determination of these metabo-lites in natural products, is of paramount importance.

Determination of triterpenoids in plant extracts is ratherdifficult, since many plants contain a vast amount of various triter-penoid compounds. Presence of isomeric triterpenoids in plantcuticular waxes renders the determination of triterpenoids evenmore difficult. Among the separation techniques, thin-layer (TLC)[1316], supercritical fluid (SFC) [17], gas (GC) [1824] and high-performance liquid (HPLC) [13,14,2333] chromatography andcapillary electrophoresis (CE) [34] have been used in their anal-ysis. GC is favorable for the separation of positional triterpenoidisomers, and coupled to flame ionizaton (FID) [18,19,2124] and

mass spectrometric (MS) detection [1820,22,24] has been widelyused for their qualitative [1820,22,24] and quantitative analysis[18,19,2124]. A big disadvantage of this technique is the need of aprechromatographic derivatization, due to the non-volatility of the

dx.doi.org/10.1016/j.chroma.2015.01.001http://www.sciencedirect.com/science/journal/00219673http://www.elsevier.com/locate/chromahttp://crossmark.crossref.org/dialog/?doi=10.1016/j.chroma.2015.01.001&domain=pdfmailto:irena.vovk@ki.sidx.doi.org/10.1016/j.chroma.2015.01.001

2 omato

chdmswaioc[itooipostsavcnso

sniedt

2

2

iehsomTewrcaFbCMf

2

fwt

30 K. Naumoska, I. Vovk / J. Chr

ompounds, which additionally prolongs the analysis. On the otherand, HPLC reduces the sample pretreatment step by avoiding theerivatization. However, triterpenoids and phytosterols lack chro-ophores, which limits the mobile phase choice and reduces theensitivity of UV detection [13,14,23,25]. Therefore, coupling HPLCith evaporative light scattering (ELSD) [24,26] or corona chargederosol (CAD) detector [27] can be suitable, since they showedncreased sensitivity compared to UV detectors for the compoundsf interest. In addition, triterpenoids can be tagged with fluores-ent groups and determined by fluorescence detector (FLD), as well28,29]. However, UVvis, as a non-selective and universal detector,s the most commonly used in HPLC. Furthermore, the mass spec-rometry (MS) detector enables identification and quantificationf compounds in real samples, and aids the structural elucidationf unknown compounds as well [13,14,26,3033]. When producton (MS2) analysis is used in the identification of the triterpenoidositional isomers, the spectra differ only by the relative intensitiesf some mass peaks, which indicates that a good chromatographiceparation prior to MS analysis is obligatory for unambiguous iden-ification of the compounds [13,14]. Although, TLC offers lowereparation efficiencies compared to GC and HPLC, it is a highlypplicable technique especially for fast screening of compounds inarious complex matrices simultaneously, since the sample purifi-ation step is usually avoided or is minimal. In addition, there areo limitations in the selection of mobile phase solvents in compari-on to HPLC. Moreover, an increase in the specificity and sensitivityf the analysis can be achieved by its coupling to tandem MS [35].To the best of our knowledge, there is a lack of methods for

imultaneous identification or determination of common planteutral triterpenoids and phytosterols. Moreover, there is scarcenformation on their presence in the cuticular waxes of various veg-tables. Therefore, the main objective of the present study was toevelop and apply a new TLCMS2 method for the analysis of theriterpenoids and phytosterols in vegetable extracts.

. Experimental

.1. Chemicals

All the solvents used in the study were at least of analyt-cal grade. Dichloromethane, chloroform, n-propanol, n-hexane,thyl acetate, acetic acid (glacial, 100%), sulfuric acid (9597%),ydrochloric acid (fuming, 37%) sodium sulfate (anhydrous), pota-ium hydroxide and 4-methoxybenzaldehyde (anisaldehyde) werebtained from Merck (Darmstadt, Germany), while HPLC gradeethanol and acetonitrile were produced by J.T. Baker (Deventer,he Netherlands). Acetone, LCMS purity acetonitrile as well as ref-rence standards for ursolic acid (90%) and -sitosterol (97%)ere purchased from SigmaAldrich (St. Louis, MO, USA). Othereference standards such as lupeol (99%), -amyrin (98.5%),-amyrin (98.5%), cycloartenol (90%), lupeol acetate (95%),ycloartenol acetate (90%), lupenone (95%), friedelin (99%)nd betulinic acid (97%) were supplied by Extrasynthse (Genay,rance), while stigmasterol (99%) was obtained from Serva Fein-iohemica (Heidelberg, Germany) and oleanolic acid (97%) fromarl Roth (Karlsruhe, Germany). Ultrapure water was supplied by ailli-Q water purification system (18 M cm) from Millipore (Bed-

ord, MA, USA).

.2. Preparation of standard solutionsStock solutions of all standards (1 mg mL1; except those ofriedelin and betulinic acid with concentration of 0.1 mg mL1)ere prepared in n-propanol and were further diluted with

he same solvent to obtain working solutions (25 g mL1). Agr. A 1381 (2015) 229238

mixture of all 13 standard solutions (MIX13, 25 g mL1; lupeol,-amyrin, -amyrin, cycloartenol, cycloartenol acetate, lupeolacetate, lupenone, friedelin, ursolic acid, oleanolic acid, betulinicacid, -sitosterol and stigmasterol) and a mixture with all thestandards except ursolic, oleanolic and betulinic acids (MIX10,25 g mL1) were prepared by mixing 1 mL of each workingsolution, evaporating the solvent under nitrogen and redissolvingthe solid residue in 1 mL of n-propanol.

2.3. Preparation of vegetable extracts and sample test solutions

The extraction of the vegetable cuticular waxes and prepa-ration of sample test solutions followed the procedure given inRef. [13]. Fresh vegetables were purchased from a local market.Fruits from zucchini (Cucurbita pepo L., Cucurbitaceae; 1845 g), egg-plant (Solanum melongena L., Solanaceae; 2775 g), tomato (Solanumlycopersicum L., Solanaceae; 4289 g) and red pepper (Capsicumannuum L., Solanaceae; 696 g) and leaves from mangold (Betavulgaris L. ssp. vulgaris var. cicla, Chenopodiaceae; 77 g), spinach(Spinacia oleracea L., Chenopodiaceae; 194 g), lettuce (Lactuca sativaL. var. capitata, Cichoriaceae; 350 g), white-colored radicchio diCastelfranco (Cichorium intybus L. var foliosum, Cichoriaceae; 368 g),radicchio Leonardo (Cichorium intybus L. var. foliosum, Cichori-aceae; 529 g), white cabbage (Brassica oleracea L. subsp. oleraceaconvar. capitata L. var. capitata L. f. alba, Brassicaceae; 488 g),red cabbage (Brassica oleracea L. subsp. oleracea convar. capitataL. var. capitata L. f. rubra, Brassicaceae; 250 g), and savoy cab-bage (Brassica oleracea L. subsp. oleracea convar. capitata (L.) Alef.var. sabauda, Brassicaceae; 157 g) were separately immersed intodichloromethane for 1 min. After addition of anhydrous sodiumsulfate (1 g) to the extract (to bind the residual water), it wasfiltered through paper filter and the filtrate was concentratedunder reduced pressure (Rotavapor, Bchi, Switzerland). The con-centrated extract was transferred to a pre-weighted plastic tube(15 mL), and the solvent was evaporated to dryness by using agentle stream of nitrogen. Dry wax residues (61370 mg) were dis-solved in chloroform to a concentration of 10 mg mL1 and 1 mLof each vegetable extract was transferred in a separate autosam-pler vial. The extract solvent was evaporated to dryness and thesolid residue was redissolved in n-propanol (1 mL) to give the finalsample test solution. A hair dryer was used to speed up the solva-tion process of the waxy residues. The sample test solutions werecooled down to room temperature and filtered through a 0.45 mMillipore Millex-HV hydrophilic poly(vinyldiene difluoride) (PVDF)membrane filter (Billerica, MA, USA).

2.4. Thin-layer chromatography

TLC was performed on the Merck 20 cm 10 cm glass-backedHPTLC silica gel 60 (Art. No. 1.05641) and HPTLC C18 RP (Art. No.1.05914) plates predeveloped with chloroformmethanol (1:1, v/v)and acetone, respectively, and dried in an oven at 110 C for 30 min.Standard solutions, MIX10, MIX13 and sample test solutions wereapplied on the plates as 8 mm (or 6 mm for TLCMS) bands, 10 mmfrom the bottom of the plates, by use of Linomat 5 (Camag, Muttenz,Switzerland). Plates used for compounds screening were devel-oped to a distance of 8 cm (in 12 min) in a horizontal developingchamber (for 20 cm 10 cm plates; Camag) using 6 mL of develop-ing solvents n-hexaneethyl acetate (5:1, v/v) for silica gel platesand acetoneacetonitrile (5:1, v/v) and ethyl acetateacetonitrile(3:2, v/v) for C18 RP HPTLC plates [13]. For each case, 10 mL of

the corresponding solvent was put in a tank for preconditioning(10 min).

As a part of the optimization of the separation for TLCMSanalysis, C18 RP HPTLC plates were developed by ethyl

omatogr. A 1381 (2015) 229238 231

abt(f

oo2v

awwm

ctao(cwa

2

(tSawuostmwitrcmscAtradtaboooIa

adew

Fig. 1. (Left side) The images of a plate, captured under white light and at 366 nmare showing separation of eight triterpenoids and two phytosterols on HPTLC C18RP plate developed with ethyl acetateacetonitrile (3:2, v/v) to a distance of 18 cm.(Right side) Schematic representation of a plate prepared for band elution by TLCMSK. Naumoska, I. Vovk / J. Chr

cetateacetonitrile (3:2, v/v) in different developing cham-ers (Camag): normal, twin trough [13], horizontal (sandwich andank configuration) and Automatic development chamber ADC2Camag). Different application positions (5 mm, 8 mm and 10 mm)rom the bottom were also tested.

Finally, TLC screening and TLCMS2 analysis were performedn 10 cm 20 cm C18 RP HPTLC plates, developed to a distancef 18 cm (in 1 h 20 min) in a normal developing chamber for0 cm 20 cm plates using 20 mL of ethyl acetateacetonitrile (3:2,/v) as a developing solvent.Anisaldehyde detection reagent was prepared by mixing glacial

cetic acid (20 mL) and methanol (170 mL). The solution was cooledith ice and water while sulfuric acid (16 mL) was added in a drop-ise manner. Subsequently, anisaldehyde (1 mL) was added to theixture [36].After developing and drying of the plates, the post-

hromatographic derivatization was performed by dippinghe plates into the anisaldehyde detection reagent for 2 s by usingn immersion device III (Camag), followed by drying with a streamf warm air and heating them at 110 C on a TLC plate heaterCamag) for 2 min (silica gel plates) or 30 s (C18 RP plates). Thehromatographic plates were captured under UV (366 nm) andhite light by using DigiStore 2 Documentation System (Camag)nd winCATS software (version 1.4.3.6336).

.5. TLCMS2 and mass spectral library

A TLCMS interface with oval elution head (4 mm 2 mm)Camag) was employed for the elution of the compounds fromhe C18 RP HPTLC plates into a LCQ ion trap (Thermo Finnigan,an Jose, CA, USA) system. Acetonitrile was used as an eluent at

flow rate of 0.3 mL min1. Acetonitrile flow (0.3 mL min1) andorking standard solution flow (5 L min1) were joined by a T-nion and directed into the MS source for direct syringe infusionf the standards. Atmospheric pressure chemical ionization (APCI)ource operated in positive ion mode was used for ionization of thearget compounds. Capillary and source heater temperatures wereaintained at 225 C and 400 C, respectively. Other parametersere as follows: sheath gas (N2) 65 arbitrary units (a.u.), auxil-

ary gas (N2) 27 a.u., discharge current 3 A, capillary voltage 18 V,ube lens offset 25 V. The mass spectra were acquired in the m/zange 300500. Compromise MS2 conditions were set for all theompounds, except for friedelin, lupenone, -sitosterol and stig-asterol (Table 1). Each sample test solution was analyzed on aeparate 10 cm 20 cm C18 RP HPTLC plate in seven equal appli-ations (30 L), which followed the application of MIX10 (20 L).fter the development, only the MIX10 track and the closest sampleest solution track were derivatized with anisaldehyde detectioneagent, while the remaining six applications were used for TLCMSnalysis (Fig. 1). Each zone with RF equal to the RF of the stan-ards (marked with a soft pencil) was eluted from the plate andransferred to the MS detector followed by first-order mass (MS)nd product ion (MS2) spectra acquisition. The interday repeata-ility of the obtained spectra was also checked. The mass spectraf the zones from sample test solutions were compared with thosef the standards (25 g mL1) acquired by direct syringe infusionr directly by elution from the undeveloped HPTLC plates (250 ng).n addition, plate background spectra at different RF values werecquired in order to avoid false positives.The NIST Mass Spectral Search Program 2.0 enabled creation of mass spectral library from the MS2 spectra of the individual stan-ards. The acquired MS2 spectra of the sample test solution zonesluted from the plates were exported to the library and comparedith the standard spectra.interface.

3. Results and discussion

3.1. TLC

The initial screening of 11 triterpenoids (triterpenols: lupeol,-amyrin, -amyrin and cycloartenol; ketones: friedelin andlupenone; acetates: cycloartenol acetate and lupeol acetate; andacids: ursolic, oleanolic and betulinic acids) and two phytos-terols (-sitosterol and stigmasterol) in the vegetable sampletest solutions was performed by three TLC methods (Fig. 2) [13],which did not enable satisfactory separation between the acids.The first method, employing HPTLC silica gel plate developedwith n-hexaneethyl acetate (5:1, v/v), enabled separation of thecompounds by their functional groups. Positional isomers werecompletely or partly separated by the other two methods, usingHPTLC C18 RP plates and two different developing solvents, i.e.acetoneacetonitrile (5:1, v/v) and ethyl acetateacetonitrile (3:2,v/v). The latter developing solvent showed to be the most appropri-ate, as it provided higher selectivity and resolution among isomers,i.e. cycloartenol and positional isomers, -amyrin and -amyrin,while lupeol, cycloartenol acetate and friedelin bands were over-lapped, which was not the case when using acetoneacetonitrile(5:1, v/v). The observed bands from the sample test solutions werecompared with those obtained from the standards by their RF andcolor (given also in Ref. [13]) after derivatization with anisaldehydedetection reagent. At this stage the compounds identity cannot beinferred with certainty. However, these experiments showed thatthe vegetable cuticular waxes can be potential sources of triter-penoids and phytosterols.

As expected, development of C18 RP HPTLC plates by ethylacetateacetonitrile (3:2, v/v) in different developing chambers

(normal, twin-trough, horizontal (sandwich and tank configura-tion), ADC2) and application position from the bottom of the plateinfluenced the RF values and the resolution between the closely

232 K. Naumoska, I. Vovk / J. Chromatogr. A 1381 (2015) 229238

Table 1Selected MS2 conditions for the studied compounds. Isolation width of m/z 1.5 and activation radio frequency of 0.25 were set for all the compounds.

Compound Parent mass (m/z) Normalized collision energy (%) Activation time (ms) Range (m/z)

Lupeol

409 35 60.000 150420

-Amyrin-AmyrinCycloartenolCycloartenol acetateLupeol acetate

Friedelin 427 45 100.000 150440Lupenone 425 55 60.000 300430-Sitosterol 397 45 60.000 150410Stigmasterol 395 35 60.000 150410

Fig. 2. Screening of triterpenoids and phytosterols in the sample test solutions obtained from the different vegetable wax extracts and documented at 366 nm. a: HPTLCsilica gel 60 plates developed in n-hexaneethyl acetate (5:1, v/v); b: HPTLC C18 RP plates developed in acetoneacetonitrile (5:1, v/v); c: HPTLC C18 RP plates developedin ethyl acetateacetonitrile (3:2, v/v). Tracks: 1 = mangold (15 L), 2 = savoy cabbage (10 L), 3 = red cabbage (11 L), 4 = MIX13 (10 L), 5 = spinach (15 L), 6 = red pepper(8 L), 7 = white-colored radicchio di Castelfranco (4 L), 8 = white cabbage (7 L), 9 = eggplant (18 L), 10 = lettuce (9 L), 11 = radicchio Leonardo (1 L), 12 = tomato (1 L),13 = zucchini (16 L).

K. Naumoska, I. Vovk / J. Chromato

Fig. 3. C18 RP HPTLC plates developed with ethyl acetateacetonitrile (3:2, v/v)in different developing chambers for 20 cm 20 cm (1), 20 cm 10 cm (49) and10 cm 10 cm plates (2, 3, 1012): normal unsaturated (1, 2) and 10 min saturated(3), ADC2 (4, 5, 6), twin trough (7, 8, 9), horizontal-tank configuration without (10)and with (11) 10 min of preconditioning, horizontal-sandwich configuration (12).D(b

es

bbo1witmaaizm

TApw

3.2. Stability of standards on the plate

The standards applied on the plate were stable for at least 14 hafter development, which enabled TLCMS analysis. On the othereveloping solvent volume: 20 mL (1), 6 mL (49), 5 mL (2, 3), 3 mL (1012). MIX13

10 L) was applied 5 mm (4, 7), 8 mm (5, 8) or 10 mm (13, 6, 912) from theottom.

luting compounds (Fig. 3). However, none of them provided theatisfactory resolution for TLCMS analysis.

To obtain more reliable qualitative results and discriminateetween the target compounds, we modified one of the methods,y turning a 20 cm 10 cm HPTLC C18 RP plate for 90 and devel-ping it with ethyl acetateacetonitrile (3:2, v/v) up to distance of8 cm (Fig. 1). The RF values of the bands of each compound (Table 2)ere similar to those reported in Ref. [13] for the method employ-

ng the same stationary phase and developing solvent. However,he bands of the standards of MIX10 were wider for the modifiedethod (cycloartenol, -amyrin, -amyrin, cycloartenol acetatend lupeol 0.3 cm) in comparison to the initial one (cycloartenol,-amyrin, -amyrin 0.2 cm, while the bands of cycloartenol

cetate and lupeol were overlapped). The danger to elute zonesn a non-selective manner (elution of the part of the neighboringone/s together with the target zone) was obvious for the initialethod where some bands were of even lower width than the

able 2verage RF values with standard deviations (SD) calculated for each studied com-ound from three HPTLC C18 RP plates developed in the normal developing chamberith ethyl acetateacetonitrile (3:2, v/v) to a distance of 18 cm.

Compound Average RF values SD (n = 3)-Sitosterol 0.15 0.01Stigmasterol 0.16 0.02Cycloartenol 0.28 0.01-Amyrin 0.30 0.02-Amyrin 0.33 0.01Cycloartenol acetate 0.40 0.01Lupeol 0.42 0.01Friedelin 0.43 0.01Lupeol acetate 0.50 0.01Lupenone 0.57 0.01gr. A 1381 (2015) 229238 233

width of the oval elution head (0.2 cm). This risk is especially pro-nounced when MS cannot aid in the differentiation of the possiblycoeluted compounds, as is the case with cycloartenol, -amyrin and-amyrin and cycloartenol acetate and lupeol (further referred to ascritical compounds). To the contrary, the modified method result-ing in wider bands for the critical compounds (0.3 cm), enabledhigher certainty that no coelution of the neighboring zone/s willoccur to the MS (especially when the central part of the bandis eluted). The obtained higher resolution between the bands incomparison to the initial method, due to the longer developing dis-tance and consequently the higher number of theoretical plates(N), enabled more reliable visual identification, as well as optimalseparation of all 10 compounds for MS analysis (Fig. 1). Regard-less of the non-selective elution, zigzag elution pattern could beimplemented for the TLCMS analysis of the tightly located zonesfrom the initial method to avoid leakage from the elution head. Onthe contrary, the modified method enabled straight elution pat-tern (each zone could be eluted from the same track) without anyleakage originating from the elution head. The method was appliedfor identification of eight triterpenoids and two phytosterols indifferent vegetable sample test solutions. Triterpenoid acids werefurther excluded from the study, since we already reported themin the same vegetable waxes. For their unambiguous identification,a TLCMS2 approach, based on the separation using HPTLC C18 RPplate and n-hexaneethyl acetate (5:1, v/v) as a developing solventwas used [15].Fig. 4. Background mass spectra acquired at different RF values from two differ-ent batches of HPTLC C18 RP plates predeveloped with acetone and developed inethyl acetateacetonitrile (3:2, v/v) and mass spectrum of -amyrin eluted fromthe corresponding plate.

234 K. Naumoska, I. Vovk / J. Chromatogr. A 1381 (2015) 229238

300

0

1000

100 397

395

-sitosterol

stigmasterol

0

100

0

100409

lupeol

lupeol actate409

0

1000

100 409-amyin

-amyrin 409

0

0

100

Re

lative

Ab

un

da

nce

409cycloartenol

cycloartenol

actate

427

409100

0

100

100 427

425

466

friedelin

lupenone

407

468

0

400350 500450300

0

100

0

100 271

285

257231339299

203217177 367 408

271285245231

217299203

353339 367313191409

0

100

0

100 217

203191 257285299177 231

353245 325161367 408

217203191

271285 353231177 299

245163327 367

381

271

257

0

100

0

100271257

285

229215

243 353203191 299313 367 408

271257285

229243215

203353

191339

299 366408313

0

100

0

100R

ela

tive

Ab

un

da

nce

409

191177203 231163 259 285 357313 369

407

337311 355369

217

200 250 300 350 4000

1000

100 243 257203161287

189 261229275

382355

297

255 283 311269

161201 241 325 339 367187

215

185215

301315

327341

191

257

245

313

250200 300 400350

250200 350 400

MS2 spectra:MS spectra:

dards

hhoafhucitdd

3

pbwdmwaonTd

m/z

Fig. 5. MS and MS2 spectra of the stan

and, derivatized friedelin (after 30 s of heating on the TLC plateeater) was stable only 12 min, after which the pale yellow colorbserved under illumination at 366 nm disappeared. Interestingly,fter reheating the plate for additional 30 s, the colored band ofriedelin appeared again, indicating a reversible reaction. Wheneated the plate for longer time (1, 1.5 or 2 min), the colored prod-ct of friedelin does not disappear for at least 25 min. However, wehose 30 s of heating for C18 RP plates on the TLC plate heater ast was optimal time for obtaining a good (bright) background andhe stability of other standards was not an issue under these con-itions. Therefore, the plates were captured immediately after theerivatization (in 12 min).

.3. Background mass signals in TLCMS

Serious difficulties in the TLCMS analysis can arise from thelate (adsorbent) background spectrum, which can be influencedy the plate alone (different adsorbent areas of the same plate), asell as solvents and solvent mixtures used for the predevelopment,evelopment or even for the elution to the MS. Some backgroundass signals can lead to suppression of the analyte signal/s. Thisas reported for TLC coupled to electrospray ionization MS whencidic solvents were used for predevelopment or/and development

f silica gel plates [37,38]. Therefore, differentiation of analyte sig-al/s from background mass signals is playing a crucial role in theLCMS analysis. We examined the influence of acetone as a pre-eveloping solvent and ethyl acetateacetonitrile (3:2, v/v) as am/z

eluted from the HPTLC C18 RP plates.

developing solvent on the HPTLC C18 RP plate background spec-trum and its impact on the analyte mass spectrum from elutionsat different RF values. The (+)APCI-MS spectra obtained from theplates predeveloped with acetone showed the mass peaks at m/z354, 368, 437, 453 and 497. The predeveloped HPTLC C18 RP plates(batch: HX229037), which were additionally developed with ethylacetateacetonitrile (3:2, v/v), resulted in MS background spec-tra containing four major mass peaks at m/z 354, 391, 419, 475.The same signals were obtained from zones at different RF valueson different plates with the same batch used for the whole study,suggesting that the same contamination was occurring each time.However, a batch-to-batch repeatability was not observed. Namely,when another batch (HX43689814) was analyzed for backgroundsignals, other major peaks at m/z 355 and 429 were repeating atdifferent RF values and different plates. Fortunately, no backgroundmass signals that could interfere with the mass signals of the tar-get compounds were observed (Fig. 4). Additionally, no difficultiesregarding the analyte/s ionization were noticed, which made thisTLCMS study possible without any special demands.

3.4. First-order mass (MS) and product ion (MS2) spectra

MS and MS2 spectra of the standards are given in Fig. 5.

Lupeol, -amyrin, -amyrin and cycloartenol standards (C30H50O,M = 426.72 g mol1) gave MS spectra with a base peak at m/z409 which was assigned to a dehydrated protonated molecule[M+HH2O]+. Additionally, in the MS spectrum of cycloartenol,

K. Naumoska, I. Vovk / J. Chromatogr. A 1381 (2015) 229238 235

F e MS2

r

at[tcsMta(cdpstdmtitpkirD

ido

ig. 6. Histograms of the proposed characteristic ions (marked by *) observed in thelative intensities, calculated from the four replications, are taken into account.

mass peak at m/z 427 was observed corresponding to the pro-onated molecule [M+H]+. Along with the base peak at m/z 427M+H]+, the friedelin standard (C30H50O, M = 426.72 g mol1) spec-rum showed also a mass peak of lower intensity at m/z 468orresponding to a protonated cluster ion [M+H+CH3CN]+. Thetandards of cycloartenol acetate and lupeol acetate (C30H52O2,

= 468.75 g mol1) showed similar and identical MS spectra asheir free forms, respectively, revealing a base peak at m/z 409,ssigned as [MCH3COO]+. The mass spectrum of lupenoneC30H48O, M = 424.7 g mol1) showed a base peak at m/z 425,orresponding to the protonated molecule [M+H]+, as well asehydrated protonated molecule [M+HH2O]+ at m/z 407 and arotonated cluster ion [M+H+CH3CN]+ at m/z 466. Standard MSpectra of -sitosterol (C29H50O, M = 414.71 g mol1) and stigmas-erol (C29H48O, M = 412.69 g mol1) revealed base peaks of theehydrated protonated molecules [M+HH2O]+ at m/z 397 and/z 395, respectively. In addition to the molecular mass informa-

ion, the developed APCI-MS method in positive ion mode providesnformation with regard to the presence of functional group inhe compound structure. Thus, mass spectra of triterpenols andhytosterols (containing hydroxyl functional groups), triterpenoidetones and triterpenoid acetates showed base peaks correspond-ng to ions assigned as [M+HH2O]+, [M+H]+ and [MCH3COO]+,espectively, which is in line with the conclusions drawn by Vaner Doelen and coworkers [31].

MS2 spectra did not show any significant interday variation

n the fragmentation patterns, when acquired in four differentays. Under the MS2 conditions, the protonated molecules [M+H]+

f friedelin (m/z 427) and lupenone (m/z 425) were fragmentedspectra of the triterpenoid compounds. For the ordinate axis, the mean values for

to product ions with m/z values of 409 ([M+HH2O]+) and 407([M+HH2O]+), respectively. The other compounds gave MS2 spec-tra of more complex nature, which cannot be interpreted in astraightforward manner. The MS2 analysis of lupeol, -amyrin,-amyrin, cycloartenol, cycloartenol acetate and lupeol acetateresulted in similar MS2 spectra containing a group of intensiveproduct ions with m/z values of 257, 271 and 285, except forcycloartenol and cycloartenol acetate. Moreover, for the MS2 spec-tra of cycloartenol, cycloartenol acetate, -amyrin and -amyrin,additional group of product ions is observed with m/z values of191, 203 and 217. For cycloartenol and cycloartenol acetate, thisgroup is dominant, while for -amyrin and -amyrin it is of lowerintensity. Lupeol and lupeol acetate have the additional commonproduct ion at m/z 215 which is of higher intensity than that atm/z 217. In comparison to the other compounds, -sitosterol andstigmasterol showed different fragmentation patterns. Histogramsof the proposed discrimination markers, denoted as characteristicions, observed in the MS2 spectra are shown in Fig. 4. Similar dataare published for the fragmentation of triterpenols [13,14,32,39]and phytosterols [32].

3.5. TLCMS2

TLCMS2 was shown to be suitable for qualitative analysis ofthe eight triterpenoids and two phytosterols. Their characteris-

tically colored bands [13] were satisfactory separated and weredivided in four groups by their migration on the TLC plate (Fig. 1).The compounds from a certain RF range gave characteristic ion(s)in the MS or MS2 spectra, by which they could be differentiated

236 K. Naumoska, I. Vovk / J. Chromatogr. A 1381 (2015) 229238

Fig. 7. Assignation of some compounds extracted from red pepper cuticular wax. (a) Parallel mass spectra comparison: the upper two spectra in each MS and MS2 comparisonsection belong to the eluted sample test solution zones from the developed HPTLC C18 RP plate, whilst the third is the spectrum of the corresponding standard and (b) headt zoned

faldtswntcbtAtaIdtwmwf

o tail comparison of the MS2 spectra obtained from the eluted sample test solutionownward pointed mass spectra, respectively.

rom the closely eluting compounds (Figs. 5 and 6). -sitosterolnd stigmasterol from the first group (RF range: 0.120.21) had theowest RF on the plate. The discrimination between them could beone relatively easy, due to their good chromatographic separa-ion, different band colors and different MS and MS2 spectra. Theecond group (RF range: 0.240.35) was consisted of cycloartenol,-amyrin and -amyrin. Despite their relatively close elution, theyere distinguished by different colors of their bands. Unfortu-ately, they could not be unambiguously identified by MS, sinceheir MS spectra showed the same base peak at m/z 409. However,ycloartenol could be differentiated from the other two triterpenolsy the presence of an additional mass peak at m/z at 427 in its spec-rum. Their MS2 spectra showed similar fragmentation patterns.s previously explained for the cycloartenol standard, the spec-rum showed dominant group of ions with m/z values of 191, 203nd 217, and a minor one with m/z values of 257, 271 and 285.n contrast, the MS2 spectra of -amyrin and -amyrin showed aominant group of ions peaking at m/z 257, 271 and 285. However,he characteristic group of ions with m/z values of 191, 203 and 217

as more intensive in the spectrum of -amyrin than that of its iso-er, -amyrin. The third group of compounds (RF range: 0.370.45)as comprised of closely eluting cycloartenol acetate, lupeol and

riedelin. However, their discrimination by mass spectrometry wass from the plate and the corresponding standards, represented by the upward and

more straightforward. Friedelin gave different MS and MS2 spectrathan the closely eluting compounds. Although cycloartenol acetateand lupeol had identical MS spectra, their MS2 spectra showeddifferences between the intensities of the characteristic groups ofions, allowing compound discrimination. Cycloartenol acetate andlupeol MS2 spectra showed the same dominant group of ions ascycloartenol and amyrins, respectively. The MS2 spectrum of lupeolshowed a characteristic product ion peak at m/z 215, which canbe observed also in the MS2 spectrum of lupeol acetate. Lupeolacetate and lupenone from the forth group (RF: 0.470.59) werewell separated from each other and from the other compounds onthe plate. Lupeol acetate MS and MS2 spectra are practically identi-cal to those of lupeol. A similar case was observed for cycloartenoland cycloartenol acetate MS2 spectra, as well. Lupenones distinc-tive MS and MS2 spectra enabled straightforward identification.

3.6. Analysis of the sample test solution compounds MS2 spectraby mass spectral libraryAlthough the modified TLC method facilitated the discrimi-nation between the target compounds compared to the initialmethod, MS2 analysis was required for identification of the com-pounds in the sample test solutions. However, without the created

K. Naumoska, I. Vovk / J. Chromatogr. A 1381 (2015) 229238 237

Table 3Triterpenoids and phytosterols identified in different sample test solutions verified by the mass spectral library. Shaded fields correspond to positively assigned compoundsin the vegetable test solutions. The assignments that were previously reported in the literature are marked with asterisk.

Vegetable -Sitosterol Stigmasterol Cycloartenol -Amyrin -Amyrin Cycloartenolacetate

Lupeol Friedelin Lupeolacetate

Lupenone

Zucchini * *Eggplant * * * * *Tomato * * * * * *Red pepper * *MangoldSpinach * *Lettuce

White-coloredradicchio diCastelfranco

Radicchio Leonardo

mtcgshzdmpdihtr

pvmAb[adogfitTt

4

oTwFwsaacFfm

[

White cabbage * * * Red cabbageSavoy cabbage

ass spectral library, built from the MS2 spectra of the standards,he identification of the compounds in the sample test solutionsould not be done clearly. As the closely eluting positional isomersave complex and similar MS2 spectra, positive assignments basedolely on the library statistical results were not made. Therefore,ead to tail comparison between the acquired sample test solutionones MS2 spectra and those of the corresponding standards wasone. This comparison mode plots the relative intensities against/z, where the first spectrum points upward, and the second oneoints downward from a common axis. Therefore, similarities andifferences in the fragmentation patterns could be seen clearer thann the parallel mass spectra comparing. The axis symmetry in theead to tail output indicates positive assignment. Example of howhe qualitative analysis of cycloartenol, -amyrin and -amyrin ined pepper test solution was performed is shown in Fig. 7.

To the best of our knowledge, this is the first report of theresence of certain triterpenoids and phytosterols in the selectedegetable waxes (Table 3). Table 3 summarizes the positive assign-ents verified by mass spectral library (head to tail comparison).

great part of the previously published data (Table 3, markedy *) was in accordance with that reported in the literature13,14,2022,4043]. The compounds that were not positivelyssigned were not present in the studied vegetable types, notetected due to the low concentrations in the sample test solutionsr not detected due to spectra averaging between the coeluting tar-et and matrix compounds (false negative). Moreover, this is therst simultaneous qualitative TLCMS2 study of the common neu-ral triterpenoids and phytosterols in plant materials. A qualitativeLCMS2 study of triterpenoid acids was also reported for the firstime by our group [15].

. Conclusions

Three TLC methods were employed for the initial screeningf triterpenoids and phytosterols in vegetable cuticular waxes.he preliminary experiments showed that the vegetable cuticularaxes could be potential sources of triterpenoids and phytosterols.or the purpose of their identification, a new TLCMS2 methodas proposed. It combined the advantages of the satisfactoryeparation of the target compounds, their different band colors,s well as the possibility of their discrimination based on MSnd MS2 spectra. Since the MS2 spectra were similar among the

losely eluting compounds, the characteristic ions were proposed.or unambiguous identification, a mass spectral library was builtrom the MS2 spectra of the target compounds. The head to tailode of comparison between the sample test solution zones and

[

[

* *

corresponding standard MS2 spectra have increased the clarityof the results obtained. This study reports the first simultaneousTLCMS2 method for the identification of the common neutraltriterpenoids and phytosterols in vegetable extracts. In additionto the molecular mass information, the developed (+)APCI-MS2

method provided information regarding to the nature of themolecular functional group, as well. This approach enabled posi-tive assignment of different triterpenoids and phytosterols in thestudied vegetable cuticular waxes, of which many are identifiedfor the first time.

Acknowledgements

The study was carried out with the financial support from theSlovenian Research Agency (P1-0005) and EN-FIST Centre of Excel-lence. The authors want to express their most sincere gratitudeto Mateja Puklavec, Andreja Starc, Dr. Vesna Glavnik and Dr. AlenAlbreht for the generous help during the experimental work, as wellto Dr. Zoran Kitanovski for the help during the preparation of themanuscript.

References

[1] T.J. Walton, Waxes, cutin and suberin, in: J.L. Harwood, J.R. Bowyer (Eds.),Methods in Plant Biochemistry. Volume 4: Lipids, Membranes and Aspects ofPhotobiology, Academic Press, London, 1990, pp. 105158.

[2] P.M. Dewick, Medicinal Natural Products: A Biosynthetic Approach, second ed.,John Wiley and Sons Ltd., Chishester, 2002.

[3] J. Liu, Pharmacology of oleanolic acid and ursolic acid, J. Ethnopharmacol. 49(1995) 5768.

[4] Z. Ovesn, A. Vachlkov, K. Horvthov, D. Tthov, Pentacyclic triterpenoicacids: new chemoprotective compounds, Neoplasma 51 (2004) 327333.

[5] P. Dzubak, M. Hajduch, D. Vydra, A. Hustova, M. Kvasnica, D. Biedermann, L.Markova, M. Urban, J. Sarek, Pharmacological activities of natural triterpenoidsand their therapeutic implications, Nat. Prod. Rep. 23 (2006) 394411.

[6] M.B.C. Gallo, M.J. Sarachine, Biological activities of lupeol, Int. J. Biomed. Pharm.Sci. 3 (2009) 4666.

[7] J. Patocka, Biologically active pentacyclic triterpenes and their current medicinesignification, J. Appl. Biomed. 1 (2003) 712.

[8] L.H. Vzquez, J. Palazon, A. Navarro-Ocana, The pentacyclic triterpenes ,-amyrins: a review of sources and biological activities, in: V. Rao (Ed.), Phyto-chemicals A Global Perspective of Their Role in Nutrition and Health, InTech,Rijeka, 2012, pp. 487502.

[9] G.F. Aragao, M.C.C. Pinheiro, P.N. Bandeira, T.L.G. Lemos, G.S. de Barros Viana,Analgesic and anti-inflammatory activities of the isomeric mixture of alpha-and beta-amyrin from Protium heptaphyllum (Aubl.) March, J. Herbal Pharma-cother. 7 (2007) 3147.

10] T.G. Tolstikova, I.V. Sorokina, G.A. Tolstikov, A.G. Tolstikov, O.B. Flekhter, Bio-logical activity and pharmacological prospects of lupane terpenoids. I. Natural

lupane derivatives, Russ. J. Bioorg. Chem. 32 (2006) 3749.

11] T.P. Carr, M.M. Ash, A.W. Brown, Cholesterol-lowering phytosterols: factorsaffecting their use and efficacy, Nutr. Diet. Suppl. 2 (2010) 5972.

12] W.H. Ling, P.J.H. Jones, Dietary phytosterols: a review of metabolism, benefitsand side effects, Life Sci. 57 (1995) 195206.

http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0220http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0225http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0230http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0235http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0240http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0245http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0250http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0255http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0260http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0265http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0270http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0275

2 omato

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

38 K. Naumoska, I. Vovk / J. Chr

13] M. Martelanc, I. Vovk, B. Simonovska, Separation and identification of somecommon isomeric plant triterpenoids by thin-layer chromatography and highperformance liquid chromatography, J. Chromatogr. A 1216 (2009) 66626670.

14] M. Martelanc, I. Vovk, B. Simonovska, Determination of three majortriterpenoids in epicuticular wax of cabbage (Brassica oleracea L.) byhigh-performance liquid chromatography with UV and mass spectrometricdetection, J. Chromatogr. A 1164 (2007) 145152.

15] K. Naumoska, B. Simonovska, A. Albreht, I. Vovk, TLC and TLCMS screening ofursolic, oleanolic and betulinic acids in plant extracts, J. Planar Chromatogr. 26(2013) 125131.

16] O.N. Pozharitskaya, S.A. Ivanova, A.N. Shikov, V.G. Makarov, Separation andquantification of terpenoids of Boswellia serrata Roxb. extract by planar chro-matography techniques (TLC and AMD), J. Sep. Sci. 29 (2006) 22452250.

17] E. Lesellier, E. Destandau, C. Grigoras, L. Fougre, C. Elfakir, Fast separation oftriterpenoids by supercritical fluid chromatography/evaporative light scatter-ing detector, J. Chromatogr. A 1268 (2012) 157165.

18] N. Stiti, S. Triki, M.-A. Hartmann, Formation of triterpenoids throughout Oleaeuropaea fruit ontogeny, Lipids 42 (2007) 5567.

19] L.P. Halinski, J. Szafranek, B.M. Szafranek, M. Goebiowski, P. Stepnowski, Chro-matographic fractionation and analysis of the main components of eggplant(Solanum melongena L.) leaf cuticular waxes, Acta Chromatogr. 21 (2009)127137.

20] S. Bauer, E. Schulte, H.-P. Thier, Composition of the surface wax from tomatoes.I. Identification of the components by GC/MS, Eur. Food Res. Technol. 219 (2004)223228.

21] S. Bauer, E. Schulte, H.-P. Thier, Composition of the surface wax from tomatoes.II. Quantification of the components at the ripe red stage and during ripening,Eur. Food Res. Technol. 219 (2004) 487491.

22] S. Bauer, E. Schulte, H.-P. Thier, Composition of the surface waxes from bellpepper and eggplant, Eur. Food Res. Technol. 220 (2005) 510.

23] S. Jger, H. Trojan, T. Kopp, M.N. Laszczyk, A. Scheffler, Pentacyclyc triterpenedistribution in various plants-rich sources for a new group of multi-potent plantextracts, Molecules 14 (2009) 20162031.

24] .P. Halinski, M. Paszkiewicz, M. Goebiowski, P. Stepnowski, The chemicalcomposition of cuticular waxes from leaves of the gboma eggplant (Solanummacrocarpon L.), J. Food Compos. Anal. 25 (2012) 7478.

25] C. Mathe, G. Culioli, P. Archier, C. Vieillescazes, High-performance liquidchromatographic analysis of triterpenoids in commercial frankincense, Chro-matographia 60 (2004) 493499.

26] B. Rhourri-Frih, P. Chaimbault, D. Dequeral, P. Andr, M. Lafosse, Investiga-tion of porous graphitic carbon for triterpenoids and natural resinous materialsanalysis by high performance liquid chromatography hyphenated to mass spec-trometry, J. Chromatogr. A 1240 (2012) 140146.

27] M. Plante, B. Bailey, C. Crafts, I.N. Acworth, Sensitive HPLC Method for Triter-penoid Analysis using Charged Aerosol Detection with Improved Resolution,Thermo Fisher Scientific, Pittcon, Chelmsford, MA, USA, 2012, March, pp. 17.28] S. Zhang, Y. Sun, Z. Sun, X. Wang, J. You, Y. Suo, Determination of triterpenicacids in fruits by a novel high performance liquid chromatography method withhigh sensitivity and specificity, Food Chem. 146 (2014) 264269.

29] H. Wu, G. Li, S. Liu, D. Liu, G. Chen, N. Hu, Y. Suo, J. You, Simultane-ous determination of six triterpenic acids in some Chinese medicinal herbs

[

[

gr. A 1381 (2015) 229238

using ultrasound-assisted dispersive liquidliquid microextraction and high-performance liquid chromatography with fluorescence detection, J. Pharm.Biomed. Anal. 107 (2015) 98107.

30] N. Snchez-vila, F. Priego-Capote, J. Ruiz-Jimnez, M.D. Luque de Cas-tro, Fast and selective determination of triterpenic compounds in oliveleaves by liquid chromatographytandem mass spectrometry with multiplereaction monitoring after microwave-assisted extraction, Talanta 78 (2009)4048.

31] G.A. Van Der Doelen, K.J. Van Den Berg, J.J. Boon, N. Shibayama, E. Ren De LaRie, W.J.L. Genuit, Analysis of fresh triterpenoid resins and aged triterpenoidvarnishes by high-performance liquid chromatographyatmospheric pressurechemical ionization (tandem) mass spectrometry, J. Chromatogr. A 809 (1998)2137.

32] W. Zarrouk, A. Carrasco-Pancorbo, A. Segura-Carretero, A. Fernndez-Gutirrez,M. Zarrouk, Exploratory characterization of the unsaponifiable fraction ofTunisian virgin olive oils by a global approach with HPLC-APCI-IT MS/MS anal-ysis, J. Agric. Food Chem. 58 (2010) 64186426.

33] B. Rhourri-Frih, P. Chaimbault, B. Claude, C. Lamy, P. Andr, M. Lafosse, Analysisof pentcyclic triterpenes by LCMS. A comparative study between APPCI andAPPI, J. Mass Spectrom. 44 (2009) 7180.

34] H.-Y. Cheung, Q.-F. Zhang, Enhanced analysis of triterpenes, flavonoids andphenolic compounds in Prunella vulgaris L. by capillary zone electrophoresiswith the addition of running buffer modifiers, J. Chromatogr. A 1213 (2008)231238.

35] J.J. Monaghan, W.E. Morden, T. Johnson, I.D. Wilson, P. Martin, Thinlayer chromatography/mass spectrometry: the advantages of tan-dem mass spectrometry, Rapid Commun. Mass Spectrom. 6 (1992)608615.

36] H. Jork, W. Funk, W. Fischer, H. Wimmer, Dnnschicht-Chromatographie.Reagenzien und Nachweismethoden, Bd. 1a, Physikalishe und chemische Nach-weismethoden: Grundlagen, Reagenzien I, VCH Verlagsgesellschaft, Weinheim,1989.

37] G.E. Morlock, Background mass signals in TLC/HPTLC-ESI-MS and practicaladvices for use of the TLCMS interface, J. Liq. Chromatogr. Relat. Technol. 37(2014) 28922914.

38] S. Smrke, I. Vovk, Comprehensive thin-layer chromatography mass spec-trometry of flavanols from Juniperus communis L. and Punica granatum L, J.Chromatogr. A 1289 (2013) 119126.

39] M. Morita, M. Shibuya, T. Kushiro, K. Masuda, Y. Ebizuka, Molecular cloning andfunctional expression of triterpene synthases from pea (Pisum sativum), Eur. J.Biochem. 267 (2000) 34533460.

40] S.D. Eigenbrode, S.K. Pillai, Neonate Plutella xylostella responses to surface waxcomponents of a resistant cabbage (Brassica oleracea), J. Chem. Ecol. 24 (1998)16111627.

41] L. Normn, M. Johnsson, H. Andersson, Y. van Gameren, P. Dutta, Plant sterolsin vegetables and fruits commonly consumed in Sweden, Eur. J. Nutr. 38 (1999)

8489.

42] V. Piironen, J. Toivo, R. Puupponen-Pimi, A.-M. Lampi, Plant sterols in vegeta-bles, fruits and berries, J. Sci. Food Agric. 83 (2003) 330337.

43] J.-H. Han, Y.-X. Yang, M.-Y. Feng, Contents of phytosterols in vegetables andfruits commonly consumed in China, Biomed. Environ. Sci. 21 (2008) 449453.

http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0280http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0285http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0290http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0295http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0300http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0305http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0310http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0315http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0320http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0325http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0330http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-9673(15)00027-8/sbref0335http://refhub.elsevier.com/S0021-967