Mq

Ra

b

h

�

�

�

�

a

ARRAA

KMSMAQ

1

ry

Mf

0h

Analytica Chimica Acta 770 (2013) 1– 6

Contents lists available at SciVerse ScienceDirect

Analytica Chimica Acta

j ourna l ho me page: www.elsev ier .com/ locate /aca

ultiple headspace-solid-phase microextraction: An application touantification of mushroom volatiles

osaria Costaa, Laura Tedonea, Selenia De Graziaa, Paola Dugoa,b, Luigi Mondelloa,b,∗

Dipartimento Farmaco-chimico, University of Messina, viale Annunziata, 98168 Messina, ItalyCentro Integrato di Ricerca (C.I.R.), Università Campus-Biomedico, Via Álvaro del Portillo, 21, 00128 Roma, Italy

i g h l i g h t s

Multiple headspace extraction-solidphase microextraction (MHS-SPME)has been applied to the analysis ofAgaricus bisporus.Mushroom flavor is characterized bythe presence of compounds with a 8-carbon atoms skeleton.Formation of 8-carbon compoundsinvolves a unique fungal biochemicalpathway.The MHS-SPME allowed to deter-mine quantitatively 5 target analytesof A. bisporus for the first time.

g r a p h i c a l a b s t r a c t

r t i c l e i n f o

rticle history:eceived 26 November 2012eceived in revised form 17 January 2013ccepted 20 January 2013vailable online 6 February 2013

eywords:

a b s t r a c t

Multiple headspace-solid phase microextraction (MHS-SPME) followed by gas chromatography/massspectrometry (GC–MS) and flame ionization detection (GC–FID) was applied to the identification andquantification of volatiles released by the mushroom Agaricus bisporus, also known as champignon. MHS-SPME allows to perform quantitative analysis of volatiles from solid matrices, free of matrix interferences.Samples analyzed were fresh mushrooms (chopped and homogenized) and mushroom-containing fooddressings. 1-Octen-3-ol, 3-octanol, 3-octanone, 1-octen-3-one and benzaldehyde were common con-

ultiple headspace extractionolid phase microextractionushroom flavor

garicus bisporusuantitative analysis

stituents of the samples analyzed. Method performance has been tested through the evaluation of limitof detection (LoD, range 0.033–0.078 ng), limit of quantification (LoQ, range 0.111–0.259 ng) and ana-lyte recovery (92.3–108.5%). The results obtained showed quantitative differences among the samples,which can be attributed to critical factors, such as the degree of cell damage upon sample preparation,that are here discussed. Considerations on the mushrooms biochemistry and on the basic principles of

sente

MHS analysis are also pre

. Introduction

Solid-phase microextraction is a well established sample prepa-ation technique that has gained an enormous success during theears, dating back to more than 20 years ago. From the pioneer

∗ Corresponding author at: Dipartimento Farmaco-chimico, University ofessina, viale Annunziata, 98168 Messina, Italy. Tel.: +39 090 6766536;

ax: +39 090 358220.E-mail address: lmondello@unime.it (L. Mondello).

003-2670/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.aca.2013.01.041

d.© 2013 Elsevier B.V. All rights reserved.

works by Pawliszyn and co-workers published in 1992, the num-ber of publications has grown exponentially up to around 1084papers, based on the use of SPME, in 2011 [1]. SPME is easy, fast,simple, convenient, and environmentally friendly. However, one ofthe features of this technique, which turns to be at the same timea drawback, is that SPME performs a non-exhaustive extraction.In SPME, the process of extraction is based on the achievement

of equilibria between sample matrix and headspace, and betweenheadspace and fiber coating. A SPME extraction is considered com-plete when the equilibria are established, although this phasedoesn’t correspond necessarily to the exhaustion of analytes from

2 Chimi

tbmaawn(iicciiipteucn

ehtmPespgyrcpdpPeifp

A

wficsitr

l

ws

aepet[

R. Costa et al. / Analytica

he sample matrix. This issue makes somehow challenging cali-ration procedures when SPME is chosen as sample preparationethodology. In order to make quantitation of SPME extracted

nalytes, a variety of calibration procedures are available to thenalyst as suggested by the manufacturer [2]. The decision ofhich approach is the most convenient (internal standard, exter-al standard or standard addition) depends upon sample matrixliquid or solid), its complexity and extraction mode (headspace ormmersion). The external standard calibration generally succeedsn the construction of calibration graphs with good linear regressionoefficients; however, this method doesn’t take into account the soalled “matrix effect” which causes target analytes to be embeddedn a complex matrix, where they establish several uncontrollablenteractions with other constituents. The internal standard methods mostly advised for simple matrices, due to the fact that in a com-lex sample, the standard added can undergo the matrix effect inhe same way as the other constituents. In order to avoid the matrixffect, for complex samples the standard addition method can besed, but even in this case, the calibration methodology is in mostases neither feasible nor reliable. The standards added and theative analytes behave differently [3].

An interesting alternative to eliminate the sample matrixffects when analyzing VOCs from solid samples is multipleeadspace extraction (MHE). This analytical approach dates backo 1970, when Suzuki et al. introduced a new method for esti-

ating occluded solvents in adhesive tapes [4]. Called “Multiplehase Equilibration,” this calibration procedure was successivelymployed by McAuliffe in the determination of hydrocarbons dis-olved in water [5]. In this last paper, McAuliffe postulated therinciples of the MHE theory, reporting equations, calibrationraphs and practical implications of the methodology. About 10ears later, Kolb dealt again with the topic through an extensiveeview focusing on MHS theoretical background and practical cal-ulations over a various range of samples (crude oil, sutures, food,harmaceuticals) [6]. Quoting Kolb et al., “MHE is in principle aynamic gas extraction procedure, but carried out stepwise, com-arable to a repeated liquid extraction in a separation funnel” [7].ractically, the same sample is subjected to a number of consecutivextractions, generally corresponding to three or four, at equal timentervals. The total peak area of the target analyte can be drawnrom the geometric progression, obtained from the consecutiveeak areas of the single extractions:

T =n∑

i=1

Ai = A1

1 − f= A1

1 − e−q′

here AT is the total peak area, A1 the analyte peak area from therst extraction, f the quotient of the geometric progression, q′ aonstant which takes into account the distribution coefficient andome instrumental parameters. From this equation, one can eas-ly understand that the two values necessary for the calculation ofhe total area are A1 and q′. The latter can be obtained from linearegression analysis of the following equation:

n Ai = −q′ · i − 1 + ln A1

hich corresponds to a y = mx + b type linear equation where thelope m = −q′.

Once obtained the AT value, the real concentration of the targetnalyte in the original matrix can be gathered from a simultaneousxternal calibration graph, constructed apart with standard com-

ounds either by direct injection or by MHS-SPME extraction. Anxtensive discussion on the MHS theoretical principles and prac-ical implications has been presented by Kolb and Ettre in 19918].

ca Acta 770 (2013) 1– 6

In the present study, the theory of multiple headspace extractionhas been applied, in combination with SPME, to the quantificationof some key compounds released by the mushroom Agaricus bis-porus. This fungus belongs to the edible mushrooms category; itis better known as “champignon” and can be commonly found onthe vegetables counter. Around A. bisporus there’s a big market,since it is widely used in the food industry (frozen and cannedmushrooms, soups, pizza, pasta dressings, aroma extracts, etc.).Literature reports 8-carbon atoms skeleton compounds, such as 3-octanone, 3-octanol, (2E)-octenol, as key compounds of mushroomflavor [9]. Previous studies about A. bisporus dealt with packag-ing and storage vs. flavor quality and biochemical pathway ofthe mushroom life cycle [10,11]. Once developed the MHS-SPMEmethod, this was applied also to some food formulations containingA. bisporus.

2. Experimental

2.1. Samples

Mushrooms belonging to the species A. bisporus were pur-chased in local grocery stores and were immediately analyzed.Upon receipt, samples were divided in two groups and subjectedto different preparation procedures: parts were added with dis-tilled water and peanut oil (2:1:1) and homogenized; parts werecoarsely chopped and then added with water and oil in the sameratio as above. Also, a commercial cream, used as food dressing andlabeled as containing 23% of A. bisporus, was purchased in a grocerystore and analyzed without any pre-treatment.

About 0.1 g of each sample were put into a 10 mL crimped vial forSPME extraction. A mix of C7–C30 n-alkanes (Supelco, Bellefonte,CA, USA) was extracted by SPME and desorbed into the GC–MS sys-tem in order to measure the experimental linear retention indices(LRIs).

Stock solutions (1000 ppm) of 1-octen-3-ol, 3-octanone, 3-octanol, benzaldehyde, 1-octen-3-one, benzyl alcohol, phenylac-etaldehyde, (2E)-octenol and 1-octanol were prepared in peanutoil and serial dilutions in the range 0.001–20 �g g−1 were extractedby multiple headspace SPME. All the standards were provided bySigma–Aldrich (St. Louis, MO, USA).

2.2. SPME conditions

SPME extraction was carried out in the headspace mode bymeans of an AOC-5000 autosampler (Shimadzu, Kyoto, Japan)hyphenated with the GC–MS system. Two different fiber coatingswere tested: a 65 �m polydimethylsiloxane/divinylbenzene, 1 cmlong; and a 50/30 �m DVB/Car/PDMS, 1 cm long, both provided bySupelco (Bellefonte, CA, USA). After SPME method development,the PDMS/DVB fiber was chosen to extract the volatile compo-nents from the mushrooms. Fiber exposure lasted 20 min at 50 ◦C,under agitation. Analytes were then desorbed for 1 min at 250 ◦Cin the GC injector in splitless mode, equipped with a 0.75 mm IDinlet liner. Multiple headspace extraction was performed throughfour consecutive extractions, with a 5 min interval between eachof them. For external standard calibration, graphs were built-upon 5 points, each corresponding to the total areas (AT) obtainedfrom the MHS-SPME extraction of standard compounds at differentconcentrations.

2.3. GC–FID analyses

For gas chromatographic separations, a Shimadzu GC-2010 Plussystem was used (Shimadzu, Kyoto, Japan). The split/splitless injec-tor was held at a temperature of 250 ◦C, and, after sampling time

Chimica Acta 770 (2013) 1– 6 3

(rtSgFfl

2

PGat2owcas

3

epRevta

Rsy

Table 1GC–MS data obtained from the analyses of mushrooms and mushroom fooddressing.

Peak no. Compound LRIlib LRIexp Similarity score

1 Benzaldehyde 960 961 962 1-Octen-3-one 973 976 943 1-Octen-3-ol 978 981 984 3-Octanone 986 985 955 3-Octanol 999 998 956 Benzyl alcohol 1040 1035 947 Phenylacetaldehyde 1045 1044 978 (2E)-Octenol 1067 1068 969 1-Octanol 1076 1072 96

LRIlib: Linear Retention Index from the mass spectral library FFNSC 2 (Shimadzu,

Fp

R. Costa et al. / Analytica

1 min) in splitless mode, a split ratio of 1:20 was applied. Car-ier gas was helium, at a constant linear velocity of 30.0 cm s−1. Allhe analyses were carried out on a 30 m × 0.25 mm i.d. × 0.25 �m dfLB-5MS column (Supelco, Bellefonte, CA, USA), temperature pro-rammed as follows: 40 ◦C at 3 ◦C min−1 to 250 ◦C, held 10 min. TheID temperature was set at 330 ◦C and gas flows were 40 mL min−1

or hydrogen and 400 mL min−1 for air, respectively. Data were col-ected by using GCsolution software (Shimadzu, Kyoto, Japan).

.4. GC–MS analyses

GC–MS analysis was performed on a Shimadzu GCMS-QP2010lus, equipped with the same column used for GC–FID analysis.as chromatographic parameters were the same as for GC–FIDnalyses. Carrier gas inlet pressure was 26.7 kPa. The ion sourceemperature was set at 220 ◦C, and the interface temperature was50 ◦C. The acquisition took place in scan mode at a scan intervalf 0.20 s within a mass range of 40–400 amu. The software usedas GCMSsolution by Shimadzu (Shimadzu, Kyoto, Japan). Identifi-

ation was performed with the support of mass spectral databasesvailable on the market plus some home-made libraries collectingpectra from food and flavor samples [12,13].

. Results and discussion

Fig. 1 shows the GC chromatograms relative to chopped, homog-nized fresh mushrooms, and a commercial food dressing. Therecision of the method was evaluated by the measurement ofSD% coming from the peak areas of three replicates of MHS-SPMExtractions on real samples, under optimized conditions. The RSD%alues fell, on average, in the range 3.0–7.8%, with an outlier regis-ered for benzyl alcohol (20.9%), which resulted to be problematics ahead will be discussed.

Compounds identified in mushroom flavor, along with Linearetention Indices (both experimental and published) and masspectral similarity score, are reported in Table 1. The GC–MS anal-sis, with the support of the LRI filter and highly reliable mass

ig. 1. GC elution profiles of different samples of mushroom flavors: (a) chopped mushroeak identity.

Kyoto, Japan).LRIexp: Linear Retention Index experimentally measured for real samples.

spectral databases, produced results in agreement with data previ-ously reported for this type of samples. Furthermore, the identityof each compound was confirmed by the injection of the standardcompounds utilized for MHS external calibration. In fact, as can beseen, the volatile composition of A. bisporus mushrooms is char-acterized by eight carbon atoms compounds, such as 1-octen-3-oland 3-octanone, and considerable amounts of compounds with thearomatic ring moiety [14].

A major consideration can be drawn from the observationof the chromatograms shown in Fig. 1: the main eight-carbonvolatile found in chopped samples was 3-octanone, while it was1-octen-3-ol in homogenized samples. This phenomenon confirmsfurther that enzymes responsible for eight-carbon compounds for-mation are stored in specific cell compartments, the disruption ofwhich makes them free, thus activating the fatty acids breakdown.Polyunsaturated fatty acids are the precursors of a wide range ofshort-chain volatiles in fungi, utilized by these organisms as natural

weapons against pests or as metabolites involved in reproduction[15]. Many studies have demonstrated that linoleic acid, apart frombeing the most abundant PUFA in A. bisporus, represents also themost common original substrate for eight-carbon compounds [16].

om; (b) homogenized mushroom; and (c) mushroom-based cream. See Table 1 for

4 R. Costa et al. / Analytica Chimica Acta 770 (2013) 1– 6

F rent tI

Api

Fuoc

dra

l

u

A

thdrqb

Ft

ig. 2. GC chromatogram expansions relative to four consecutive extractions of diffeII: third extraction; and IV: fourth extraction.

s mentioned in the introductory section, MHS-SPME is a stepwiserocedure which subjects to a number i of consecutive extractions

n the same sample.The visual consequence of such a methodology is represented in

ig. 2, where it can be easily realized how peak areas/heights grad-ally decrease between successive extractions. Based on the theoryf MHS, in the present application each sample was extracted fouronsecutive times, with a 5 min interval in between of them.

For each target compound to be quantified, the GC peak areaerived from every SPME-GC analysis was registered and loga-ithmically plotted vs. the number of extractions (see Fig. 3). Thisllowed to obtain the q′ value, based on the equation

n Ai = −q′ · i − 1 + ln A1

seful for the calculation of AT:

T =n∑

i=1

Ai = A1

1 − f= A1

1 − e−q′

Correlation coefficients higher than 0.99% were observed for allhe components determined, with the exception of benzyl alco-ol and phenylacetaldehyde. Unfortunately, these two compoundsidn’t show an exponential decrease in the peak areas, neither ineal samples nor in calibration analyses. It must be emphasized that

′ is a constant that depends on the distribution kinetics of analytesetween the solid matrix and the gaseous phase:

− dcdt = kc (1st order kinetics)where c = concentration; t = time.

ig. 3. Two examples of linear plots (ln Ai vs. i − 1) which describe the coefficient ofhe geometric progression f for 3-octanone and 1-octen-3-one.

ypes of samples showing decreasing peaks. I: first extraction; II: second extraction;

In other words, the concentration of analyte in the headspacedecreases proportionally to the time increase, but only in equilib-rium conditions.

Sometimes, lack of linearity can be caused by a wrong amount ofsample: a too low amount doesn’t allow to reach even the limit ofdetection; on the other hand, an oversized sample would requirean endless number of extractions (in this case, microextractions)to produce a noticeable peak area decrease. Finally, the chemicalnature of the fiber coating should be taken in consideration, sincesome compounds show a very high affinity toward some particularcoatings, where they get stuck, making tough even fiber clean-up.

All the results relative to the MHS-SPME-GC analysis arereported in Table 2. As can be seen from the table, f values rangedfrom 0.458 to 0.906, demonstrating that the MHS-SPME methoddeveloped is efficient and reliable. In fact, according to previouspapers [3,17], MHS-SPME can be considered feasible only for f-values within the range 0.4–0.95.

Table 2 reports also method validation data. Limits of detec-tion (LoD) and of quantification (LoQ) were measured accordingto Martínez-Urunela et al. [18]: first extractions of the mostdiluted standard solution were considered for signal-to-noise ratiosof 3 and 10, respectively. Recoveries were obtained by spikingmushroom samples with 10 ppm of standard and comparing theconcentration of the added standard to that coming from the cali-bration equation. Recoveries ranged from 92.35% to 108.5%.

Quantitation of flavor compounds by MHS-SPME highlighteda predominance of benzaldehyde in homogenized mushrooms(4.13 �g g−1). However, this compound was not detected in thefood dressing, probably because the industrial treatment leads toa loss of this volatile or transformation into other compounds; or,alternatively, artificial addition of chemicals smelling as mushroom(8 carbon compounds), instead of natural A. bisporus could havebeen done by the producer.

Prior to the MHS-SPME application, SPME method developmentwas carried out, therefore tuning all the different parameters affect-ing the microextraction process. Sample size, water–oil addition,fiber coating, pre-incubation, extraction time and temperature, agi-tation, were tested for method optimization.

The choice of using peanut oil as vehicle for mushrooms and

standard compounds arised from the evaluation of different mate-rials as potential solvents. This choice is critical to the reliability ofthe results relative to the composition of mushroom flavor that ina very susceptible way can drastically change. Beyond the damage

R. Costa et al. / Analytica Chimica Acta 770 (2013) 1– 6 5

Table 2MHS-SPME data measured for quantification of key compounds in mushroom flavor.

ln f f AT (�g g−1) LoD (ng) S N−1 = 3 LoQ (ng) S N−1 = 10 Recovery (%)(10 ppm spiked)

1-Octen-3-ol 0.052 0.174 103.2Chopped −0.169 0.845 869,140 0.75Homogenized −0.598 0.549 3,287,658 3.30Cream −0.199 0.819 3,215,724 3.22

3-Octanone 0.042 0.140 108.5Chopped −0.780 0.458 3,358,165 3.34Homogenized −0.306 0.736 2,033,276 2.01Cream −0.449 0.638 199,948 0.18

3-Octanol 0.054 0.179 93.8Chopped −0.481 0.618 319,030 0.19Homogenized −0.212 0.809 94,915 0.07Cream −0.206 0.813 580,798 0.32

Benzaldehyde 0.078 0.259 92.3Chopped −0.197 0.821 2,793,831 1.42Homogenized −0.122 0.885 4,137,168 4.13

1-Octen-3-one 0.033 0.111 97.3Chopped −0.246 0.782 67,047 0.03Homogenized −0.752 0.471 340,552 0.30Cream −0.099 0.906 650,877 0.61

lwtlpTococ

fpso(ir

iMsawf

Ffis

evel, also the type of solvent added may play an important role:ater addition seems to work as enzymatic catalyst, accelerating

he biochemical reactions. Non-polar inorganic solvents will over-oad the fiber coating. White mineral oil and peanut oil gave the besterformance, both in terms of solubility of chemicals and inertness.he latter was evaluated by GC running repeatedly SPME extractsf these oils: no peaks could be detected, apart from some long-hain free fatty acids in the last part of the chromatograms. Peanutil was chosen because this was an ingredient of the mushroomream samples investigated.

As above mentioned, relevant quantitative fluctuations wereound for 1-octen-3-ol when fresh mushrooms were differentlyre-treated (0.75 and 3.30 �g g−1, in chopped and homogenizedamples, respectively). It seems reasonable that from 1-octen-3-l breakdown other 8-carbon compounds are formed: 3-octanone3.34 vs. 2.01 �g g−1) and 3-octanol (0.19 vs. 0.07 �g g−1) werendeed more abundant exactly in chopped samples, balancing theeduced presence of 1-octen-3-ol.

The total areas (AT) reported in Table 2 were interpolatednto the calibration equations obtained for target analytes by

HS-SPME extraction of reference standard compounds. Fig. 4

hows a plot collecting all the calibration curves with their rel-tive regression equations and correlation coefficients, whichere in the range 0.997–1.000. For each standard compound,

our consecutive SPME extractions were carried out at every

ig. 4. Calibration graphs obtained for mushroom flavor key compounds, quanti-ed by MHS-SPME extraction. Linear regression coefficients and equations are alsohown.

single concentration tested. The AT obtained from each concen-tration was used to build-up the calibration graphs shown inFig. 4.

4. Concluding remarks

Although the mushroom A. bisporus is the most traded culti-vated species worldwide, there are no studies dealing with thequantification of its volatiles extracted by SPME. The multipleheadspace-SPME method here presented has demonstrated tobe suitable for the direct determination of flavor constituentsfrom A. bisporus. However, it must be emphasized that a fun-damental prerequisite for a successful MHS-SPME application isworking in standardized conditions of high precision and accu-racy. In this case, particular care has been dedicated to somecritical variables, such as sample volume, sample pre-treatment,choice of solvents and time spent between two consecutive extrac-tions.

MHS-SPME allowed to characterize and quantify mushroomflavor key-compounds directly, without any pre-treatment. The fvalues obtained confirmed that the linearity of the method wasvery good, making it feasible to the scope of this study. Further-more, MHS-SPME resulted to be a valid tool for those who need toquantitate volatiles in solid matrices. This preliminary work willinvolve in the future the MHS-SPME quantification of differenttypes of mushrooms collected from the wild in the Mediterraneanarea.

Acknowledgments

Authors wish to thank Shimadzu Corps. and Sigma-Aldrich/Supelco for constantly supporting their research work.

References

[1] Scifinder Scholar, © 2012 American Chemical Society.[2] Supelco Bulletin, 929 (2001) 1–8.[3] M.T. Tena, J.D. Carrillo, Trends Anal. Chem. 26 (2007) 206–214.

[4] M. Suzuki, S. Tsuge, T. Takeuchi, Anal. Chem. 42 (1970) 1705–1708.[5] C. McAuliffe, Chem. Tech. 1 (1971) 46–51.[6] B. Kolb, Chromatographia 15 (1982) 587–594.[7] B. Kolb, P. Pospisil, M. Auer, Chromatographia 19 (1984) 113–122.[8] B. Kolb, L.S. Ettre, Chromatographia 32 (1991) 505–513.

6 Chimi

[

[[

[

[

[[

8724–8732.[17] E. Serrano, J. Beltrán, F. Hernández, J. Chromatogr. A 1216 (2009)

R. Costa et al. / Analytica

[9] T. Leguijt, D. Yuksel, R. van der Vuurst de Vries, M. Eillebrecht, N. Muskens, H.van der Valk, M. Sanders, T. de Rijk, H. Wichers, in: D. Royse (Ed.), MushroomBiology and Mushroom Products, Pennsylvania State University, UniversityPark, PA, USA, 1996, pp. 515–523.

10] C. Guillaume, I. Schwab, E. Gastaldi, N. Gontard, Innov. Food Sci. Emer. 11 (2010)690–696.

11] P.F.S. Byrne, P.J. Brennan, J. Gen. Microbiol. 89 (1975) 245–255.12] FFNSC 2 – Flavour and Fragrance Natural and Synthetic Compounds, mass

spectral database, Shimadzu Corps., Japan, 2011.13] R.P. Adams, Identification of Essential Oil Components by Gas Chromatogra-

phy/Mass Spectrometry, 4th ed., Allured, Carol Stream (IL), 2007.

[

ca Acta 770 (2013) 1– 6

14] E. Combet, J. Handerson, D.C. Eastwood, K.S. Burton, J. Agric. Food Chem. 57(2009) 3709–3717.

15] E. Blee, Trends Plant Sci. 7 (2002) 315–321.16] R.R. Simon, K.M. Phillips, R.L. Horst, I.C. Munro, J. Agric. Food Chem. 59 (2011)

127–133.18] A. Martínez-Urunela, J.M. González-Sáiz, C. Pizarro, J. Chromatogr. A 1089

(2005) 31–38.