Food Chemistry 271 (2019) 639–649

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Characterization of chemical compounds susceptible to be extracted from T

cork by the wine using GC-MS and 1H NMR metabolomic approaches
⁎
Joana Pintoa, , Ana Sofia Oliveiraa, Paulo Lopesb, Isabel Roseirab, Miguel Cabralb,
⁎
Maria de Lourdes Bastosa, Paula Guedes de Pinhoa,
a
UCIBIO@REQUIMTE/Laboratório de Toxicologia, Departamento de Ciências Biológicas, Faculdade de Farmácia, Universidade do Porto, Rua Jorge Viterbo Ferreira,
228, 4050-313 Porto, Portugal
b
Amorim & Irmãos, S. A, Rua dos Corticeiros 850, 4536-904 Santa Maria de Lamas, Vila da Feira, Portugal

A R T I C LE I N FO A B S T R A C T

Chemical compounds studied in this article: This work presents a metabolomics study of cork by gas chromatography-mass spectrometry (GC-MS) and 1H
Pyrogallol (PubChem CID: 1057) nuclear magnetic resonance (NMR) spectroscopy to characterize compounds susceptible to be extracted from
trans-Squalene (PubChem CID: 638072) cork by the wine in an attempt to find a relationship between the content of these compounds and the geo-
Sitost-4-en-3-one (PubChem CID: 5484202) graphical origin of cork. Cork from eleven geographical regions was studied, five from Portugal and six from
Friedelin (PubChem CID: 91472)
Spain. Unsupervised pattern recognition techniques unveiled three main clusters of regions according to their
Camphene (PubChem CID: 6616)
chemical similarity but not related with geographical proximity. Nineteen compounds were found to be re-
o-Cymene (PubChem CID: 10703)
trans-3-Pinanone (PubChem CID: 11038) sponsible for the clusters, including terpenes (trans-squalene, friedelin, camphene, trans-3-pinanone, 1-terpinen-
1-Terpinen-4-ol (PubChem CID: 11230) 4-ol, two sesquiterpenes), polyphenols (vescalagin, castalagin), among others (pyrogallol, glucosan, sitost-4-en-
Quinic acid (PubChem CID: 6508) 3-one, o-cymene, quinic acid, five unknowns). These preliminary results unveiled the potential for a more ef-
Vescalagin (PubChem CID: 5458626) ficient selection of cork planks for stoppers production based on the compounds susceptible to be extracted from
Castalagin (PubChem CID: 168165) cork by the wine.
Keywords:
Volatile compounds
Polyphenols
GC-MS
NMR spectroscopy
Metabolomics
Cork
Geographical origin

1. Introduction 1996a, 1996b; Santos, Pinto, Silvestre, & Neto, 2010; Sousa, Pinto,
Silvestre, & Neto, 2006), detecting mainly aliphatic alcohols, fatty
For several centuries, cork has been the traditional choice to seal a acids, sterols, triterpenes, monoterpenes, phenols and polyphenols.
wine bottle, protecting its qualities and allowing it to develop over These compounds can be extracted from cork by wine model solution in
time. The chemical composition of cork from Quercus suber L. consists of bottle, as demonstrated for polyphenols by Azevedo et al. (2014),
suberin, lignin, polysaccharides, extractives and other minor compo- showing their potential contribution to wine colour, flavour, as-
nents (Pereira, 2007). The extractives are low or medium molecular tringency and bitterness. Recently, it was also demonstrated that cork
weight compounds not chemically linked to its main cellular structure phenolic compounds can react with wine components ((+)-catechin
that can be extracted by solvents without affecting the structural and and malvidin-3-O-glucoside) yielding a new family of ellagitannin-de-
mechanical properties of cork (Pereira, 2007). Several reports have rived compounds (corklins) (Azevedo et al., 2017). This knowledge is of
been focused in the identification of cork extractives (Cadahía, Conde, paramount importance for wine industry, since it provides the possi-
Fernández de Simón, & García-Vallejo, 1998; Castola et al., 2005; bility to select natural cork stoppers based on the concentration of
Conde, García-Vallejo, & Cadahía, 1999a, 1999b; Moreira, Lopes, phenolic compounds thus positively influencing the chemical wine
Cabral, & Guedes de Pinho, 2016; Rocha, Delgadillo, & Ferrer Correia, development in bottle during aging.

⁎
Corresponding authors.
E-mail addresses: jipinto@ff.up.pt (J. Pinto), pguedes@ff.up.pt (P. Guedes de Pinho).

https://doi.org/10.1016/j.foodchem.2018.07.222
Received 18 December 2017; Received in revised form 24 April 2018; Accepted 31 July 2018
Available online 01 August 2018
0308-8146/ © 2018 Elsevier Ltd. All rights reserved.
J. Pinto et al. Food Chemistry 271 (2019) 639–649

The natural heterogeneity of cork chemical composition has been 2.3. Extraction of compounds from cork stopper granules
addressed by some authors in terms of its inter- and intra-tree varia-
bility (Cadahía et al., 1998; Pereira, 1988) and geographical origin Ultrasound-assisted extraction (Chemat et al., 2017) using methanol
(Boudaoud & Eveleigh, 2003; Conde et al., 1999a, 1999b; Lopes et al., was performed for extraction of semi-volatile compounds, and wine
2001). Inter-tree variability was observed for the levels of extractives model solution (12% ethanol, 5 g/L of tartaric acid at pH 3.2) for ex-
and suberin, whereas the compounds responsible for intra-tree varia- traction of volatile compounds and other major compounds. This solid-
bility were mainly suberin and lignin (Pereira, 1988). Regarding geo- liquid extraction procedure was selected after comparison with Soxhlet
graphical origin, differences in the contents of fatty acids and tri- extraction and maceration during 1–2 days at room temperature (re-
terpenes were found between cork from three Spain regions (Andalucía, sults not shown), since it had the advantage to extract the same com-
Extremadura and Cataluña) (Conde et al., 1999a, 1999b), though no pounds using a smaller volume of solvent and in a short period of time.
relationship was observed between geographical proximity and che- Briefly, 1.40 g of cork granules were mixed with 50 mL of methanol or
mical similarity. Another study (Lopes et al., 2001) reported differences wine model solution and placed in the ultrasound bath during 1 h. The
in suberin and carbohydrate contents between cork from twelve dif- cork granules from each region were extracted in quintuplicate. Quality
ferent sites in southern Portugal, not related with geographical proxi- control (QC) samples were used to monitor intra- and inter-day re-
mity but significantly dependent on the cork quality. The comparison of producibility. These samples were prepared as a pool of one replica
cork from different countries (Spain, Portugal and Morocco) was only from each region and divided into several aliquots.
reported by Boudaoud and Eveleigh (2003) showing a volatile signature
able to differentiate the geographical origin based on the levels of 2.4. Analysis of methanol extracts of cork granules by direct injection GC-
benzaldehyde, 2-ethylhexanol, acetic acid, vanillic acid, vanillin, cam- MS
phor, among other volatiles. The classification of cork samples ac-
cording to their geographical origin would benefit the cork industry Methanol extracts were evaporated under a nitrogen stream, re-
through implementation of standard quality control procedures, and suspended in a smaller volume of methanol (concentrated 4 times) and
origin certification to increase the visibility and competitivity of a 2 μL were injected in GC-MS. A 436-GC model (Bruker Daltonics)
certain geographical region. coupled to a EVOQ triple quadrupole (TQ) mass spectrometer (Bruker
This work presents a metabolomics study of cork extracts (methanol Daltonics) with a fused silica (Rxi-5Sil MS) capillary column
and wine model solution) by gas chromatography coupled to mass (30 m × 0.25 mm internal diameter × 0.25 μm; Restek Corporation,
spectrometry (GC-MS) and 1H nuclear magnetic resonance (NMR) U.S., Bellefonte, Pennsylvania) was used with the following conditions:
spectroscopy to characterize compounds susceptible to be extracted by oven temperature was held at 40 °C for 1 min, followed by increasing in
the wine and relate them to the different geographical origins of cork. an increment of 5.0 °C/min from 40 to 250 °C (5 min) and from 250 to
300 °C (0 min); injector and detector temperatures were 280 °C and
2. Materials and methods 270 °C, respectively; high purity helium C-60 (Gasin, Portugal) as car-
rier gas at a constant flow rate of 1.0 mL/min; electron impact (EI)
2.1. Chemicals mode at 70 eV; data acquisition in full scan mode and a 35–600 m/z
mass range with a scan time of 250 ms. The QC samples were repeatedly
1,4-Cineole (98.5%), α-pinene (99%), α-terpinene (95%), α-terpi- analysed under the same conditions on every five samples.
neol (98.5%), β-pinene (99%), β-sitosterol (90%), benzaldehyde The identification of semi-volatile compounds was achieved by
(99.5%), betulin (98%), (+)-camphor (95%), decanal (95%), ethyl comparing the Kovats retention indices (RI) determined for each com-
heptanoate (99%), ethyl hexanoate (99.5%), ethyl nonanoate (97%), pound in sample with the RI described in the literature and by com-
eucalyptol (99%), friedelin (95%), gallic acid (97.5%), heptanal (95%), paring the MS fragmentation with the mass spectra present in the NIST
hexanal (95%), L-borneol (99%), limonene (97%), lupeol (90%), non- 14 spectral database. When standard compounds were available, the
anal (95%), octanal (99%), quinic acid (98%), sitost-4-en-3-one, va- retention time and mass spectra obtained from sample was compared
nillin (99%) were supplied by Sigma-Aldrich, Inc. (Steinheim, with the standard compounds injected at the same conditions.
Germany). Castalagin (99.6%) and vescalagin (95.9%) were extracted
from cork, purified by preparative HPLC, identified by HPLC-DAD/ESI- 2.5. Analysis of wine model solution extracts of cork granules by GC-MS
MS and kindly provided by Faculty of Sciences, University of Porto and 1H NMR spectroscopy
(Porto, Portugal). Methanol (99.9%) and ethanol (99.9%) were sup-
plied by Sigma-Aldrich, Inc. (Steinheim, Germany) and CARLO ERBA Considering the wine model solution extracts, each one was divided
Reagents (Val de Reuil, France), respectively. Deuterium oxide (99.9%) into two aliquots: one was immediately analysed for profiling of volatile
was supplied by Euriso-top (Saint-Aubin, France). Ultrapure water was compounds by headspace solid-phase microextraction (HS-SPME) ana-
obtained from a Milli-Q Millipore purification system (Millipore, lysis and the other aliquot was lyophilized and stored at −20 °C for 1H
Billerica, Massachusetts, USA). NMR analysis.

2.2. Geographical origins and characteristics of cork 2.5.1. Analysis of wine model solution extracts of cork granules by GC-MS
For analysis of volatile organic compounds, model solution extracts
Cork planks of Quercus suber L. were sampled from eleven geo- (8 mL) were placed in a vial of 20 mL and compounds were extracted by
graphical regions, five from Portugal and six from Spain, as illustrated HS-SPME with a 50/30 μm divinylbenzene/carboxen/poly-
in Fig. S1. These regions were arbitrarily selected, and cork stoppers dimethylsiloxane (DVB/CAR/PDMS) fiber (Supelco Inc., Bellefonte,
were produced by Amorim & Irmãos, SA (Santa Maria de Lamas, Pennsylvania), using a Combi-PAL autosampler (Varian Pal
Portugal). Firstly, ten random cork planks from each geographical re- Autosampler, Switzerland). Samples were incubated (45 °C, 5 min)
gion were selected for stoppers production following conventional followed by extraction (45 °C, 30 min) under continuous stirring
procedures: 1) cork planks were boiled; 2) planks were selected ac- (250 rpm). After extraction, the thermal desorption of the analytes into
cording to thickness, porosity and appearance; and 3) selected planks the GC injector took place during 6 min at 250 °C. A 436-GC model
were punched with a machine to extract the cylindrical stoppers. Two (Bruker Daltonics) coupled to a SCION single quadrupole (SQ) mass
hundred and fifty cork stoppers were milled in a Retsch crossbeater mill spectrometer (Bruker Daltonics) was used with the same column, oven
SK1 (Haan, Germany), and the granulometric fraction of 16–35 mesh temperature and carrier gas flow conditions mentioned above (Section
was used for analysis. 2.4). The injector and detector temperature was 250 °C, EI mode at

640
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Fig. 1. Representative a) GC-MS chromatogram of semi-volatile compounds extracted from cork by methanol, b) HS-SPME-GC-MS chromatogram of volatile
compounds extracted from cork by wine model solution and c) 1H NMR spectrum of other major compounds extracted from cork by wine model solution.

641
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Table 1
List of semi-volatile (GC-MS) and volatile (HS-SPME-GC-MS) compounds extracted from cork by methanol and wine model solution, respectively.
Compound CAS RT (min) RIa Reported RIb Most abundant ions (m/z) Identification method

GC-MS profiling of semi-volatile compounds extracted from cork by methanol

Alkenes
1-Pentacosene 16980-85-1 22.63 2490 2492 43/55/69/83/97 NIST 14
1-Heptacosene 15306-27-1 26.29 2694 2693 43/57/69/83/97 NIST 14

Carbohydrate
Glucosan 498-07-7 12.01 1489 1487 57/60/73 NIST 14

Fatty alcohols
1-Eicosanol 629-96-9 20.30 2287 2281 43/55/57/69/83/97 NIST 14
1-Docosanol 661-19-8 22.87 2507 2493 43/55/69/83/97/111 NIST 14

Fatty acids
n-Hexadecanoic acid 57-10-3 17.22 1957 1968 43/60/73 NIST 14
cis-9,cis-12-Octadecadienoic acid 60-33-3 18.87 2129 2183 41/55/67/81/95 NIST 14
cis-9-Octadecenoic acid 112-80-1 18.92 2134 2175 41/55/69/83 NIST 14
n-Octadecanoic acid 57-11-4 19.14 2157 2172 43/55/57/60/73/129 NIST 14
Eicosanoic acid 506-30-9 21.02 2357 2362 43/55/57/73/129 NIST 14
Docosanoic acid 112-85-6 23.70 2558 2566 43/57/73/129/185/340 NIST 14

Glycerolipids
Glycerol 1-hexadecanoate 542-44-9 22.94 2511 2519 43/57/98/239 NIST 14
Glycerol 1-octadecanoate 123-94-4 26.73 2712 2681 43/57/98/134/267 NIST 14

Phenols and derivatives

Catechol 120-80-9 8.07 1203 1205 64/110 NIST 14
Pyrogallol 87-66-1 10.45 1370 1386 52/80/108/126 NIST 14
Vanillin 121-33-5 10.83 1398 1404 151/152 STD
trans-Coniferyl alcohol 32811-40-8 14.94 1737 1743 91/124/137/180 NIST 14

Sterols
Stigmastan-3,5-diene 214,164 34.35 3071 3040 43/81/105/147/396 NIST 14
β-Sitosterol 83-46-5 37.19 3299 3200 43/55/81/107/414 STD
Sitost-4-en-3-one 1058-61-3 39.23 3428 3447 124/229/412 STD

Triterpenes
trans-Squalene 111-02-4 29.34 2811 2832 41/69/81 NIST 14
Lupen-3-one 1617-70-5 38.07 3357 3384 81/95/109/121/205/424 NIST 14
Lupeol 545-47-1 38.53 3387 3270 81/95/107/121/189/207 STD
Friedelin 559-74-0 40.85 3515 2858 55/69/81/95/109/125 STD
Betulin 473-98-3 45.65 3713 3761 95/135/189/207/411 STD

Unknowns
Un 1 – 11.34 1437 89/136 –
Un 2 – 12.39 1520 – 75/101/116/129 –
Un 3 – 13.52 1611 – 43/60/71/112 –
Un 4 – 16.38 1873 – 43/91/103/119/131/222 –
Un 5 – 18.46 2085 – 43/55/69/83/97 –
Un 6 – 23.20 2527 – 43/57/69/82/97 –
Un 7 – 33.53 3009 – 91/105/121/147/363 –
Un 8 – 33.85 3033 – 55/93/121/163/190 –
Un 9 – 34.08 3051 – 43/55/81/105/119/143 –
Un 10 – 34.72 3101 – 55/95/163/220 –
Un 11 – 35.04 3128 – 69/93/121/135/161/381 –
Un 12 – 36.36 3236 – 69/93/121/135/161/381 –
Un 13 – 36.92 3278 – 81/134/189 –
Un 14 – 38.70 3398 – 55/95/119/134/259 –
Un 15 – 39.68 3453 – 55/93/121/163/207 –
Un 16 – 39.82 3461 – 67/79/93/121/161/189 –
Un 17 – 42.19 3575 – 43/55/79/107/149/177 –
Un 18 – 42.99 3612 – 69/81/95/107/133/203 –
Un 19 – 43.94 3649 – 55/69/95/109/123 –
Un 20 – 44.85 3685 – 55/67/81/95/119/147 –
Un 21 – 49.76 3836 – 67/81/95/137 –

HS-SPME-GC-MS profiling of volatile compounds extracted from cork by wine model solution
Aldehydes
Hexanal 66-25-1 4.82 803 800 41/44/56 STD
Heptanal 111-71-7 7.36 902 901 41/55/70 STD
Benzaldehyde 100-52-7 9.07 961 962 51/77/105/106 STD
Octanal 124-13-0 10.32 1003 1003 43/56/84 STD
Nonanal 124-19-6 13.38 1104 1104 41/57/70/98 STD
Decanal 112-31-2 16.33 1205 1206 43/57/70/82 STD

Benzenoids
o-Cymene 527-84-4 10.96 1024 1022 91/119/134 NIST 14
Naphthalene 91-20-3 15.70 1183 1185 128 NIST 14

Esters
(continued on next page)

642
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Table 1 (continued)

Compound CAS RT (min) RIa Reported RIb Most abundant ions (m/z) Identification method

Ethyl hexanoate 123-66-0 10.19 999 1000 43/60/88/99 STD

Ethyl heptanoate 106-30-9 13.15 1096 1097 88/101/113 STD
Ethyl nonanoate 123-29-5 18.83 1294 1296 88/101 STD
Fenchyl acetate 13851-11-1 16.66 1216 1223 43/81/93/121/136 NIST 14
Isobornyl acetate 125-12-2 18.60 1286 1286 43/95/121/136 NIST 14
Ester 1 – 18.00 1265 – 88/101 –

Monoterpenes
α-Pinene 80-56-8 8.25 933 937 93/121/136 STD
Camphene 79-92-5 8.73 949 952 79/93/121/136 NIST 14
β-Pinene 80-56-8 9.55 977 979 93/121/136 STD
1,4-Cineole 470-67-7 10.68 1015 1016 43/71/111/154 STD
α-Terpinene 99-86-5 10.73 1017 1017 93/121/136 STD
Limonene 5989-54-8 11.11 1029 1031 68/93/121/136 STD
Eucalyptol 470-82-6 11.19 1032 1032 43/81/108/139/154 STD
Terpinolene 586-62-9 12.80 1085 1088 93/121/136 NIST 14
Fenchone 1195-79-5 12.88 1087 1096 69/81/152 NIST 14
Fenchol 1632-73-1 13.79 1118 1113 69/80/81 NIST 14
α-Campholenal 4501-58-0 14.03 1126 1125 67/93/108 NIST 14
(+)-Camphor 464-49-3 14.61 1146 1143 41/81/95/108/152 STD
trans-β-Terpineol 7299-40-3 14.86 1154 1161 43/71/93/121/136 NIST 14
trans-3-Pinanone 547-60-4 15.01 1159 1160 41/55/69/83/95 NIST 14
Isoborneol 124-76-5 15.10 1163 1157 95/110/121/136 NIST 14
L-Borneol 464-45-9 15.34 1171 1166 95/110 STD
cis-3-Pinanone 15358-88-0 15.46 1175 1173 41/55/69/83 NIST 14
1-Terpinen-4-ol 562-74-3 15.61 1180 1177 55 / 71 / 93 / 111 NIST 14
α-Terpineol 98-55-5 16.01 1195 1189 71/93/111/136/154 STD
Monoterpene 1 – 15.23 1167 – 70/95/108/121/136 –

Sesquiterpenes
α-Copaene 3856-25-5 21.01 1376 1376 105/119/161/204 NIST 14
D-Longifolene 475-20-7 21.90 1411 1405 55/94/161/189/204 NIST 14
β-Cadinene 523-47-7 24.57 1518 1518 161/189/204 NIST 14
L-Calamenene 483-77-2 24.66 1522 1523 159/202 NIST 14
Eremophila ketone 158930-41-7 25.74 1568 1567 81/108/121 NIST 14
Sesquiterpene 1 – 22.02 1415 – 43/111/137/207/222 –
Sesquiterpene 2 22.15 1421 – 41/79/147/161/204 –
Sesquiterpene 3 – 22.77 1445 – 43/95/137/164/222 –
Sesquiterpene 4 – 23.69 1482 – 91/107/121/161/204 –
Sesquiterpene 5 – 26.01 1579 – 79/94/107/147/164/204 –

Unknowns
Un 22 – 7.93 922 – 133/151/179 –
Un 23 – 18.66 1288 – 43/57/67/82/123 –
Un 25 – 24.15 1501 – 57/205/220 –

RT: retention time; STD: standard compound; NIST 14: National Institute of Standards and Technology standard reference database, version 2.2, 2014.
a
Kovats retention indices (RI) determined using a commercial hydrocarbon mixture (C6–C20).
b
RI reported in literature.

70 eV, data acquisition in full scan mode and a 40–250 m/z mass range experiments (heteronuclear single quantum coherence, HSQC; and total
with a scan time of 500 ms. The QC sample was repeatedly analysed as correlation spectroscopy, TOCSY) and comparison with FooDB
mentioned above (Section 2.4). (Wishart, 2011) and Biological Magnetic Resonance Bank (Ulrich et al.,
Volatile organic compounds were identified as described above for 2008) databases.
semi-volatile compounds.
2.6. Statistical analysis
2.5.2. Analysis of wine model solution extracts of cork granules by 1H NMR
spectroscopy The GC-MS data was converted to CDF file format and pre-processed
For NMR analysis, model solution extracts (5 mL) were lyophilized using MZmine-2.24 (Pluskal, Castillo, Villar-Briones, & Orešič, 2010).
and resuspended in 700 μL (concentrated 6 times) deuterium oxide Pre-processing steps of methanol extracts consisted in crop filtering (m/
containing 0.1% Na+/3-trimethylsilylpropionate (TSP), followed by z 40–500 and RT 6.0–51.0 min), peak detection (noise level 1 × 105),
centrifugation at 16,060×g during 5 min. Supernatant (600 μL) was chromatogram deconvolution (peak range 0.02–0.3 min, baseline level
transferred into a 5 mm NMR tube. NMR spectra were recorded, at 2 × 105) and alignment (m/z tolerance 0.2, RT tolerance 0.05 min). For
26.85 °C (300 K), on a Bruker Avance III HD 600 MHz spectrometer model solution extracts, the pre-processing steps were as follows: crop
equipped with a cryoprobe. 1D 1H NMR pulse sequence (noesygppr1d) filtering (m/z 40–250 and RT 4.77–29.5 min), peak detection (noise
was recorded with a 1.82 s acquisition time, 2 s relaxation delay, level 1 × 104), chromatogram deconvolution (peak range
100 ms mixing time, 128 transients, 32 k complex data points, 0.02–0.3 min, baseline level 2 × 103) and alignment (m/z tolerance 0.2,
9009.01 Hz spectral width and suppression of water signal. Spectral RT tolerance 0.05 min). Artefact peaks from the chromatographic
processing consisted in zero filling to 64 k points, multiplication by a column and SPME fiber (e.g., cyclosiloxanes, siloxanes and phthalates)
1.0 Hz exponential line-broadening function, phase and baseline were manually removed. Finally, the final data matrix (11 sam-
manual correction, and chemical shifts referenced internally to TSP at ples × 1667 variables for methanol extracts and 11 samples × 450
δ = 0.0 ppm. Peak identification was carried out with basis on 2D variables for wine model solution extracts) was normalized by total

643
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Table 2
List of other major compounds (1H NMR) extracted from cork by wine model solution.
Compound CAS δ 1H ppm (multiplicity)/δ 13
C ppma Identification method

Alcohol
D-Arabitol 87-99-0 3.57 (dd)/73.40, 3.66 (m)/65.82, 3.75 (m)/74.35, 3.83 (dd)/65.81, 3.925 (m)/73.07 BMRB

Carboxylic acid
Quinic acid 77-95-2 1.92 (dd)/43.05, 2.05 (ddd)/39.62, 2.10 (d)/39.68, 2.15 (ddd)/43.12, 2.17 (ddd), 3.56 (dd)/77.74, 4.04 (dt)/69.26, 4.17 STD
(dd)/72.86

Polyphenols
Castalagin 24312-00-3 4.23 (d), 5.0 (br)/66.53, 5.08 (d)/69.02, 5.15 (t)/71.64, 5.19 (br)/76.35, 5.58 (d)/73.75, 5.76 (d)/68.37, 6.79 (s)/109.99, STD
6.85 (s)/111.24, 6.99 (s)/112.02
Vescalagin 36001-47-5 4.25 (d), 4.92 (br), 4.95 (br), 4.97 (br), 5.20 (t)/95.05, 5.39 (s)/80.02, 5.60 (d)/73.75, 6.77 (s)/109.76, 6.89 (s)/111.53, 6.98 STD
(s)/112.02

Phenolic acid
Gallic acid 149-91-7 7.13 (s)/112.85 STD

Unknowns
Un 26 – 3.25 (s)/56.26 –
Un 27 – 6.43 (s)/108.74 –
Un 28 – 7.32 (s)/111.53 –

a
600 MHz 1H and 13C NMR assignments of compounds (100% D2O with 0.1% TSP, average pH = 7.4). STD: standard compound; BMRB: Biological Magnetic
Resonance Data Bank; s: singlet; d: doublet; dd: doublet of doublets; t: triplet; tt: triplet of triplets; q: quartet; m: multiplet.

area and scaled to Pareto. The full resolution 1H NMR spectra were benzenoids, esters, monoterpenes and sesquiterpenes. Despite the vo-
subjected to pre-processing, consisting in the exclusion of the residual latile composition of cork has been subject of several studies (Moreira
ethanol signals (1.14–1.20 and 3.61–3.66 ppm) and tartaric acid et al., 2016; Rocha et al., 1996a, 1996b), nine compounds were iden-
(4.43–4.85 ppm), alignment using the recursive segment-wise peak tified in this work for the first time, according to retention indices and
alignment method (Veselkov et al., 2009) in R 3.3.3 software (R NIST 14 spectra comparisons, namely o-cymene, ethyl heptanoate, α-
Development Core Team, 2008), normalization of the final matrix (11 terpinene, fenchyl acetate, isobornyl acetate, α-campholenal, D-long-
samples × 41,275 variables) by total spectral area and scaled to Pareto. ifolene, L-calamenene and eremophila ketone. Finally, the profile of
Multivariate analysis of GC-MS and NMR data matrices was per- other major compounds susceptible to be extracted from cork by wine
formed using unsupervised pattern recognition techniques, namely model solution was analysed by 1H NMR spectroscopy being the first, to
principal component analysis (PCA) and hierarchical cluster analysis our knowledge, performed using this technique. The standard 1H NMR
(HCA). PCA was applied with a default seven-fold internal cross vali- spectrum is dominated mainly by the resonances from 5 compounds
dation in SIMCA 13.0.3 (Umetrics, Sweden). The HCA dendrograms (Fig. 1c, Table 2), namely quinic acid, arabitol, vescalagin, castalagin
were calculated using the Ward’s method, in R 3.3.3 software (R and gallic acid. The possible extraction of arabitol from cork by wine
Development Core Team, 2008). PCA loadings plots provided in- model solution was here identified for the first time, according to the
formation related to the variables (compounds detected by GC-MS or comparison with the 1H NMR spectrum of the standard compound in
NMR) responsible for the separation of the clusters observed in the BMRB database (Ulrich et al., 2008). Despite a perfect match between
binary scores plots defined by the principal components (PC1 vs. PC2) NMR spectra of pure compounds and cork extracts was found, a pos-
according to their geographical origin. The bar plots illustrating the sible contribution of other molecules (e.g., hydroxycinnamic acids,
differences in the compound levels (expressed as the mean and standard gallic acid derivatives) containing these compounds in their structure
deviation of the normalized peak area considering the five replicas) for cannot be ruled out.
each geographical region were plotted in GraphPad Prism 6 (USA). For PCA and HCA were applied to the profile of compounds susceptible
some compounds, the Pearson correlation coefficient was computed in to be extracted from cork (semi-volatiles and volatiles by GC-MS, and
R 3.3.3 software (R Development Core Team, 2008). other major compounds by 1H NMR) with the aim to find differences in
the levels of compounds from eleven geographical origins. Fig. 2 shows
the results obtained for semi-volatile compounds (Table 1) detected by
3. Results and discussion
GC-MS. The PCA scores plot (Fig. 2a) and cluster dendrogram (Fig. 2b)
show two main clusters, one comprising Douro, Córdoba I, Évora and
The profile of semi-volatile, volatile and other major compounds
Toledo and another cluster comprising the remaining geographical re-
susceptible to be extracted from cork by the wine are represented in
gions. The PC1 was the most important for discrimination of sample
Fig. 1. The profiling of semi-volatile compounds extracted from cork by
classes accounting for 80.1% of total variance. The semi-volatile com-
methanol by GC-MS (Fig. 1a, Table 1) enabled the detection of 46
pounds responsible for this discrimination were interpreted in PCA
compounds from which 21 (48% of the total detected compounds) were
loadings plot (Fig. S2a) unveiling variations in pyrogallol, glucosan,
not identified (Uni, i = 1, 2, n). The compound classes comprised al-
trans-squalene, sitost-4-en-3-one, friedelin and two unknown com-
kenes, one carbohydrate, fatty alcohols, fatty acids, glycerolipids,
pounds at RT 11.34 and 13.52 min, as illustrated in bar plots in Fig. 2c.
phenols and derivatives, steroids and triterpenes. Most of these com-
The volatile profile (Table 1) unveiled a poorer discrimination of
pounds have been described in literature as important components of
geographical regions compared with the semi-volatile profile, as can be
cork (Conde et al., 1999a, 1999b; Sousa et al., 2006), though five
seen by the PCA scores plot (Fig. 3a) and the lower R2X value of PC1
compounds were identified in this work for the first time, according to
(50.1% of total variance). In cluster dendrogram (Fig. 3b), the samples
retention indices and NIST 14 spectra comparisons, namely pyrogallol,
from Seville, Douro and Toledo stand out from the remaining origins
glucosan, glycerol 1-hexadecanoate and glycerol 1-octadecanoate. The
with sesquiterpene 1 and 3 (Fig. 3c, Fig. S2b) mainly characterizing this
profiling of volatile compounds extracted from cork by wine model
discrimination, despite a contribution of other compounds such as
solution by HS-SMPE-GC-MS (Fig. 1b, Table 1) resulted in 47 com-
camphene, o-cymene, trans-3-pinanone and 1-terpinen-4-ol.
pounds from which only 3 (6% of the total detected compounds) were
For 1H NMR profile of other major compounds (Table 2) extracted
not identified. These compounds included several aldehydes,

644
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Fig. 2. a) PCA scores scatter plot and b) HCA dendrogram obtained through comparison of semi-volatile compounds (GC-MS) extracted from cork from different
origins by methanol. c) Bar charts of compounds changing between geographical origins (expressed in average and standard deviation of quintuplicates).
Geographical origins were ordered according to their similarity in the HCA dendrogram: Douro, Córdoba I, Évora, Toledo, Cádiz, Algarve, Alcácer do
Sal, Badajoz, Castelo Branco, Seville and Córdoba II.

645
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Fig. 3. a) PCA scores scatter plot and b) HCA dendrogram obtained through comparison of volatile compounds (HS-SPME-GC-MS) extracted from cork from different
origins by wine model solution. c) Bar charts of compounds changing between geographical origins (expressed in average and standard deviation of quintuplicates).
Geographical origins were ordered according to their similarity in the HCA dendrogram: Seville, Douro, Toledo, Algarve, Badajoz, Évora, Córdoba I,
Alcácer do Sal, Córdoba II, Cádiz, Castelo Branco.

from cork, the discrimination of geographical origins in PCA scores plot foods and beverages (Tadtong, Kamkaen, Watthanachaiyingcharoen, &
(PC1 accounting for 84.5% of total variance) (Fig. 4a) and cluster Ruangrungsi, 2015; Wishart, 2011) and may contribute positively for
dendrogram (Fig. 4b) was very similar to the one obtained by profiling wine composition. Pyrogallol, which was found in higher levels in
of semi-volatile compounds, only with exception of Douro which was Cádiz, Algarve, Alcácer do Sal and Badajoz cork (Fig. 2c), belongs to the
here grouped with the remaining geographical regions. The compounds class of phenolic compounds synthesized in plants (Kocaçalişkan, Talan,
mainly responsible for the differences between origins (Fig. S2c, & Terzi, 2006), being also found in the composition of coffee, beer and
Fig. 4c) were quinic acid, vescalagin, castalagin and three unknown cocoa powder (Wishart, 2011). The antimicrobial activity of this com-
resonances, one in the carbohydrate region (3.25 ppm (s)) and two in pound was already studied in bacteria and fungi unveiling antibacterial
the polyphenol region (6.43 ppm (s) and 7.32 ppm (s)). Interestingly, an effects at concentrations of 5 and 10 mM (Kocaçalişkan et al., 2006).
inverse relationship between quinic acid and the remaining compounds Glucosan, a dehydrated glucose containing a ketal functional group,
(Fig. 4c) was observed, unveiling a negative correlation with high appeared in higher levels in Badajoz’ cork. This semi-volatile anhydride
correlation coefficients (r ranging from −0.82 to −0.90) and sig- can be formed from sugars through a complex series of reactions known
nificances (p ranging from 9.9 × 10−15 to 4.9 × 10−21). as caramelization (Huber, McDonald, & BeMiller, 2005). Squalene was
Most of the compounds susceptible to be extracted from cork stop- found in higher levels in cork from Seville, this triterpene is also present
pers by the wine, and found in this study related with different geo- in olive oil composition and is a key intermediate in cholesterol bio-
graphical origins, have been previously described in the composition of synthesis pathway (Smith, 2000). This compound has been described as

646
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Fig. 4. a) PCA scores scatter plot and b) HCA dendrogram obtained through comparison of other major compounds (1H NMR) extracted from cork from different
origins by wine model solution. c) Bar charts of compounds changing between geographical origins (expressed in average and standard deviation of quintuplicates).
Geographical origins were ordered according to their similarity in the HCA dendrogram: Córdoba I, Évora, Toledo, Douro, Córdoba II, Castelo Branco,
Seville, Cádiz, Alcácer do Sal, Algarve and Badajoz.

an important dietary cancer chemopreventive agent, possibly by acting sesquiterpenes (three isoprene units) and one benzenoid unveiled
as antioxidant and modulating enzyme activities (Smith, 2000). Sitost- changes between geographical regions (Fig. 3c). Monoterpenes and
4-en-3-one is a sterol compound that has been found in significative sesquiterpenes are constituents of essential oil produced by several
amounts (1%) in cork (Castola et al., 2005). In this study, higher levels plants and have been applied as flavours, fragrances and spices in
of this compound were observed in cork from two Portuguese regions perfumery and cosmetic products and food additives (Guimarães,
(Algarve and Alcácer do Sal). Friedelin is the triterpene found in higher Serafini, & Quintans-Júnior, 2014). Typical odour descriptors of these
abundance in cork (Pereira, 2007) and has been suggested as having compounds include camphor, herbal, woody and spicy (Wishart, 2011).
potent anti-inflammatory, analgesic and antipyretic activities Several biological properties have been attributed to terpenes including
(Antonisamy, Duraipandiyan, & Ignacimuthu, 2011). Cork is the main anticancer, antimicrobial, antifungal, antiviral, antihyperglycemic, an-
natural source of this compound, which has led to an increased interest algesic, anti-inflammatory and antiparasitic (Guimarães et al., 2014).
to develop a straightforward procedure for its extraction from cork by- Camphene, a bicyclic monoterpene also found in the composition of
products (black condensate) (Pires, Aroso, Silva, Mano, & Reis, 2011). essential oils (Tadtong et al., 2015), showed similar levels for most cork
The lowest levels of friedelin were found in cork from Douro, Córdoba I, samples of the geographical regions with exception of Seville and
Évora and Toledo. Douro, for which it was present in lower levels. trans-3-Pinanone, which
Considering the volatile compounds extracted from cork by wine is also present in herbs, spices, spearmint and roman camomile
model solution, few monoterpenes (two isoprene units) and (Wishart, 2011), was found in higher amounts in Algarve region. 1-

647
J. Pinto et al. Food Chemistry 271 (2019) 639–649

Terpinen-4-ol showed higher levels in Badajoz cork. A recent study “Contribution of cork stoppers to the chemical and sensory properties of
suggested that 1-terpinen-4-ol induces apoptosis in colorectal cancer bottled wine” co-financed by the European Regional Development Fund
cells through generation of reactive oxygen species (Nakayama et al., (FEDER) through the Operational Programme for Competitiveness and
2017). In addition, two unknown sesquiterpenes (C15 skeleton) ex- Internationalisation (COMPETE 2020), Portugal, and UID/MULTI/
hibited a very similar profile characterized by considerably higher le- 04378/2013 – POCI/01/0145/FEDER/07728. We acknowledge Dr
vels in Seville, Douro and Toledo compared with the remaining geo- Sofia Reis from Faculty of Sciences, University of Porto, for providing
graphical origins. Finally, o-cymene, identified in cork for the first time the vescalagin and castalagin standards.
in this work, was found in higher levels in Algarve, Badajoz and Cór-
doba II. This compound is one of the most abundant components of Conflicts of interest
eucalyptus oil (Tadtong et al., 2015).
The 1H NMR profile of compounds extracted from cork by wine None.
model solution revealed differences in the levels of quinic acid, vesca-
lagin, castalagin and three unknown resonances between geographical Appendix A. Supplementary data
regions (Fig. 4c). Quinic acid was found in higher levels in cork from
Córdoba I, Évora, Toledo and Douro regions. This compound is also Supplementary data associated with this article can be found, in the
present in coffee beans and other plant sources and was first identified online version, at https://doi.org/10.1016/j.foodchem.2018.07.222.
in cork extracts by Santos et al. (2010), though its quantification was
not possible due to overlapping with other peaks in high performance References
liquid chromatography (HPLC) chromatogram. Vescalagin and casta-
lagin are hydrolyzable tannins known to be released from wood into the Antonisamy, P., Duraipandiyan, V., & Ignacimuthu, S. (2011). Anti-inflammatory, an-
wine during maturation and ageing in oak barrels (García-Estévez, algesic and antipyretic effects of friedelin isolated from Azima tetracantha Lam. in
mouse and rat models. Journal of Pharmacy and Pharmacology, 63(8), 1070–1077.
Escribano-Bailón, Rivas-Gonzalo, & Alcalde-Eon, 2010). These com- Azevedo, J., Fernandes, I., Lopes, P., Roseira, I., Cabral, M., Mateus, N., & Freitas, V.
pounds are associated with wine colour and the consumer perception of (2014). Migration of phenolic compounds from different cork stoppers to wine model
astringency through interaction and/or precipitation of salivary pro- solutions: Antioxidant and biological relevance. European Food Research and
Technology, 239(6), 951–960.
teins in the mouth (Bajec & Pickering, 2008). The cluster of Córdoba I, Azevedo, J., Fernandes, A., Oliveira, J., Brás, N. F., Reis, S., Lopes, P., ... de Freitas, V.
Évora and Toledo showed the lowest levels of vescalagin and castalagin (2017). Reactivity of cork extracts with (+)-catechin and malvidin-3-O-glucoside in
whereas the cluster of Cádiz, Alcácer do Sal, Algarve and Badajoz wine model solutions: Identification of a new family of ellagitannin-derived com-
pounds (corklins). Journal of Agricultural and Food Chemistry, 65(39), 8714–8726.
showed the highest levels. In addition, three unknown resonances (Un
Bajec, M. R., & Pickering, G. J. (2008). Astringency: Mechanisms and perception. Critical
26, Un 27 and Un 28) showed a similar tendency for lower levels in the Reviews in Food Science and Nutrition, 48(9), 858–875.
first cluster and higher levels in the last one. The Un 27 (6.43 ppm, s) Boudaoud, N., & Eveleigh, L. (2003). A new approach to the characterization of volatile
signatures of cork wine stoppers. Journal of Agricultural and Food Chemistry, 51,
and 28 (7.32 ppm, s) resonances may arise from the aromatic protons of
1530–1533.
other polyphenols. Remarkably, an inverse relationship between the Cadahía, E., Conde, E., Fernández de Simón, B., & García-Vallejo, M. C. (1998). Changes
levels of quinic acid and the remaining compounds (vescalagin, casta- in tannic composition of reproduction cork Quercus suber throughout industrial
lagin, Un 26, Un 27 and Un 28) was clearly observed, although no processing. Journal of Agricultural and Food Chemistry, 46(6), 2332–2336.
Castola, V., Marongiu, B., Bighelli, A., Floris, C., Laï, A., & Casanova, J. (2005).
explanation may be advanced at this stage. Extractives of cork (Quercus suber L.): Chemical composition of dichloromethane and
supercritical CO2 extracts. Industrial Crops and Products, 21(1), 65–69.
4. Conclusions Chemat, F., Rombaut, N., Sicaire, A. G., Meullemiestre, A., Fabiano-Tixier, A. S., & Abert-
Vian, M. (2017). Ultrasound assisted extraction of food and natural products.
Mechanisms, techniques, combinations, protocols and applications. A review.
This study demonstrates the ability of GC-MS and 1H NMR spec- Ultrasonics Sonochemistry, 34, 540–560.
troscopy metabolomics to classify cork according to the chemical si- Conde, E., García-Vallejo, M. C., & Cadahía, E. (1999a). Waxes composition of re-
production cork from Quercus suber and its variability throughout the industrial
milarity of compounds extracted by methanol and wine model solution processing. Wood Science and Technology, 33(3), 229–244.
(12% ethanol, 5 g/L tartaric acid at pH 3.2). The most relevant classes Conde, E., García-Vallejo, M. C., & Cadahía, E. (1999b). Waxes composition of Quercus
of compounds extracted from cork were sterols, triterpenes, mono- suber reproduction cork from different Spanish provenances. Wood Science and
Technology, 33(4), 271–283.
terpenes, sesquiterpenes and polyphenols. These compounds may po- García-Estévez, I., Escribano-Bailón, M. T., Rivas-Gonzalo, J. C., & Alcalde-Eon, C. (2010).
sitively influence wine composition due to their contribution to aroma, Development of a fractionation method for the detection and identification of oak
colour and astringency, but also due to their potential beneficial health ellagitannins in red wines. Analytica Chimica Acta, 660(1–2), 171–176.
Guimarães, A. G., Serafini, M. R., & Quintans-Júnior, L. J. (2014). Terpenes and deriva-
effects.
tives as a new perspective for pain treatment: A patent review. Expert Opinion on
The classification of cork from different geographical origins un- Therapeutic Patents, 24(3), 243–265.
veiled three main clusters of regions differing in the levels of 19 com- Huber, K. C., McDonald, A., & BeMiller, J. N. (2005). Carbohydrate chemistry. In Y. H.
pounds (in a total of 102 detected compounds), namely pyrogallol, Hui (Ed.). Handbook of food science, technology, and engineering. Boca Raton: Taylor &
Francis Group.
glucosan, trans-squalene, sitost-4-en-3-one, friedelin, camphene, trans- Kocaçalişkan, I., Talan, I., & Terzi, I. (2006). Antimicrobial activity of catechol and
3-pinanone, 1-terpinen-4-ol, two sesquiterpenes, o-cymene, quinic acid, pyrogallol as allelochemicals. Zeitschrift Fur Naturforschung – Section C Journal of
vescalagin, castalagin and five unknown compounds. The differences in Biosciences, 61(9–10), 639–642.
Lopes, M. H., Barros, A. S., Pascoal Neto, C., Rutledge, D., Delgadillo, I., & Gil, A. M.
the levels of these compounds were not related with geographical (2001). Variability of cork from portuguese Quercus suber studied by solid-state 13C-
proximity of Quercus suber L. Thus, other factors such as different tree NMR and FTIR spectroscopies. Biopolymers – Biospectroscopy Section, 62(5), 268–277.
ages, edaphooclimatic conditions, among others, may have an influence Moreira, N., Lopes, P., Cabral, M., & Guedes de Pinho, P. (2016). HS-SPME/GC-MS
methodologies for the analysis of volatile compounds in cork material. European Food
in the discrimination of samples in addition to geographical region. Research and Technology, 242(4), 457–466.
These preliminary results show potential for a more effective selection Nakayama, K., Murata, S., Ito, H., Iwasaki, K., Villareal, M. O., Zheng, Y. W., ... Ohkohchi,
of cork planks for stoppers production taking into account the profile of N. (2017). Terpinen-4-ol inhibits colorectal cancer growth via reactive oxygen spe-
cies. Oncology Letters, 14(2), 2015–2024.
compounds susceptible to be extracted from cork stoppers by the wine. Pereira, H. (1988). Chemical composition and variability of cork from Quercus suber L.
Future studies should entail a larger cohort of training samples and Wood Science and Technology, 22(3), 211–218.
include a validation set for prediction purposes. Pereira, H. (2007). The chemical composition of cork. Cork: Biology, production and uses
(pp. 55–99). (1st ed.). Amsterdam: Elsevier.
Pires, R. A., Aroso, I., Silva, S. P., Mano, J. F., & Reis, R. L. (2011). Isolation of friedelin
Acknowledgements from black condensate of cork. Natural Product Communications, 6(11), 1577–1579.
Pluskal, T., Castillo, S., Villar-Briones, A., & Orešič, M. (2010). MZmine 2: Modular fra-
This work was supported by the CorkPlus 3310 project – mework for processing, visualizing, and analyzing mass spectrometry-based

648
J. Pinto et al. Food Chemistry 271 (2019) 639–649

molecular profile data. BMC Bioinformatics, 11(395), 1–11. Sousa, A. F., Pinto, P. C. R. O., Silvestre, A. J. D., & Neto, C. P. (2006). Triterpenic and
R Development Core Team (2008). R: A language and environment for statistical computing. other lipophilic components from industrial cork byproducts. Journal of Agricultural
Vienna, Austria: R Foundation for Statistical Computing. and Food Chemistry, 54(18), 6888–6893.
Rocha, S., Delgadillo, I., & Ferrer Correia, A. J. (1996a). GC–MS study of volatiles of Tadtong, S., Kamkaen, N., Watthanachaiyingcharoen, R., & Ruangrungsi, N. (2015).
normal and microbiologically attacked cork from Quercus suber L. Journal of Chemical components of four essential oils in aromatherapy recipe. Natural Product
Agricultural and Food Chemistry, 44(3), 865–871. Communications, 10(6), 1091–1092.
Rocha, S., Delgadillo, I., & Ferrer Correia, A. J. (1996b). Improvement of the volatile Ulrich, E. L., Akutsu, H., Doreleijers, J. F., Harano, Y., Ioannidis, Y. E., Lin, J., ... Markley,
components of cork from Quercus suber L. by an autoclaving procedure. Journal of J. L. (2008). BioMagResBank. Nucleic Acids Research, 36(Suppl. 1), D402–D408.
Agricultural and Food Chemistry, 44, 872–876. Veselkov, K. A., Lindon, J. C., Ebbels, T. M. D., Crockford, D., Volynkin, V. V., Holmes, E.,
Santos, S. A. O., Pinto, P. C. R. O., Silvestre, A. J. D., & Neto, C. P. (2010). Chemical ... Nicholson, J. K. (2009). Recursive segment-wise peak alignment of biological 1H
composition and antioxidant activity of phenolic extracts of cork from Quercus suber NMR spectra for improved metabolic biomarker recovery. Analytical Chemistry, 81(1),
L. Industrial Crops and Products, 31(3), 521–526. 56–66.
Smith, T. J. (2000). Squalene: Potential chemopreventive agent. Expert Opinion on Wishart, D. (2011). FooDB version 1.0. Retrieved July 4, 2017, from http://www.
Investigational Drugs, 9(8), 1841–1848. foodb.ca/.

649