Available online at www.sciencedirect.

com

Chemistry and Physics of Lipids 151 (2008) 6976

13

C NMR, GC and HPLC characterization of lipid components

of the salted and dried mullet (Mugil cephalus) roe bottarga
P. Scano a, , A. Rosa b , F. Cesare Marincola a , E. Locci a ,
M.P. Melis b , M.A. Dess` b , A. Lai a
b

a Dipartimento di Scienze Chimiche, Universit`

a di Cagliari, Cagliari, Italy
Dipartimento di Biologia Sperimentale, Universit`a di Cagliari, Cagliari, Italy

Received 5 June 2007; received in revised form 27 September 2007; accepted 2 October 2007
Available online 7 October 2007

Abstract
13

C NMR spectroscopy, in conjunction with HPLC and GC techniques, has been used to study the molecular composition of lipids extracted
from commercial products of bottarga. To this goal, both the saponifiable and unsaponifiable fractions of lipid extracts were also examined by 13 C
NMR. Among the major lipid classes wax esters (WE) showed a concentration of more than 50 mol%, triacylglycerols (TAG) and phospholipids
(PL) represented a minor fraction. Concentrations up to 29 mol% of free fatty acids (FFA) were found. The most represented fatty alcohol was
16:0 that accounted for more than 50%, among fatty acids the most represented were 16:1 n-7, 22:6 n-3, 18:1 n-9, 16:0, and 20:5 n-3, in particular
the n-3 polyunsaturated fatty acids (PUFA) averaged 40 mg/g of the edible portion. 13 C NMR spectroscopy put in evidence that cholesterol was
present in its free and esterified forms and its total content was measured as ca. 10 mg/g of the edible portion.
2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Fish roe; Bottarga; NMR; Lipids; Wax esters; n-3 PUFA

1. Introduction
A wide spectrum of fish roe products are consumed throughout the world. Among them, mullet (Mugil cephalus) roe is
regarded as a delicacy. The salted and dried products are known
as karasumi in Japanese and bottarga in Italian (Bledsoe
et al., 2003). Among the Mediterranean countries, Sardinia
(Italy) has a long tradition in making a high quality bottarga
that is becoming increasingly popular in the international market.
Bottarga is the final product of a number of treatments on
the ovaries of mullets. To obtain a good bottarga is an art that
requires a great deal of skills; nowadays different procedures
exist, from the more traditional to the industrial ones; in the
latter drying is conducted in rooms with controlled humidity
and temperature. The final product can be sold as whole ovaries

Corresponding author at: Dipartimento di Scienze Chimiche, Universit`

a di
Cagliari, Cittadella Universitaria di Monserrato, S.S. 554 Bivio per Sestu, 09042,
Monserrato, Cagliari, Italy. Tel.: +39 070 675 4391; fax: +39 070 675 4388.
E-mail address: scano@unica.it (P. Scano).

0009-3084/$ see front matter 2007 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.chemphyslip.2007.10.001

under vacuum packaging, or grated in jar. Bottarga has an amber

colour and its unique chewy mouth feel is due to the peculiar
lipid composition, rich in WE (Bledsoe et al., 2003).
To date, a limited number of investigations have been performed on the lipid composition of fresh and/or salted and dried
mullet roes. The proximate lipid content of fresh mullet roes
was estimated to be 13.7%, and in the salted derivates, because
of the drying process, this value was observed to increase to
23.7% (Lu et al., 1979). The lipid composition of fresh ovaries
of mullets caught in the Gulf of Mexico was extensively studied by liquid and gas chromatographic methods (Iyengar and
Schlenk, 1967). It was found that the extracted lipids exhibit a
high portion of WE (ca. 6070%), that follow the general rule
of marine WE (Nevenzel, 1969; Joh et al., 1995), i.e. are composed by saturated and unsaturated fatty acids, esterified mainly
to saturated and monounsaturated fatty alcohols of C14 to C18
chain-length. In addition, TAG, PL, cholesterol (Cho), cholesterol esters (CE), and free fatty acids and alcohols were found
in the lipid extracts (Iyengar and Schlenk, 1967). Moreover, the
fatty acids composition of salted mullet roe was reported by Lu
et al. (1979), and recently, the fatty acids and the fatty alcohols
composition of only the WE fraction in commercial prepara-

70

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

tions of bottarga has been studied by GC (Bernasconi et al.,

2007).
To our knowledge, an exhaustive study of the lipid composition of salted and dried mullet roes is lacking from the literature.
To this goal, we used the high resolution 13 C NMR spectroscopy
together with HPLC and GC techniques to investigate the lipid
extracts of different commercial samples of bottarga produced
in Sardinia. The NMR technique was used to obtain semiquantitative and multi-component information about the lipid
composition of the extracted oil, avoiding invasive manipulations of the sample. The potentiality of this technique in the
study of the molecular components of complex lipid systems
has been successfully demonstrated (Casu et al., 1991; Sacchi
et al., 1993; Aursand et al., 1993, 2007; Pollesello et al., 1996;
Scano et al., 1999, 2005, 2006; Siddiqui et al., 2003; Gribbestad
et al., 2005; Falch et al., 2006), and recently biochemical changes
in cod gonads during storage were studied by 1 H and 13 C NMR
(Falch et al., 2007).
The GC and HPLC analyses of the unsaponifiable and
saponifiable fractions of the extracted oil provided detailed
information on the individual fatty acids, fatty alcohols and
cholesterol in bottarga. In parallel, both fractions were also analyzed by NMR, in order to support, besides the GC data, the
assignment of resonances in the NMR spectrum of the pure lipid
matrix.
2. Materials and methods
2.1. Samples
Commercial products of grated mullet roe (bottarga di muggine) manufactured by two different companies located in
Sardinia (Italy) were acquired at a local supermarket. Ingredients
were reported in the label as mullet roe and salt.
2.2. Chemicals
All solvents used, of the highest available purity, were
purchased from Merck (Darmstadt, Germany). Deuterated
chloroform (CDCl3 ), l--phosphatidylcholine, oleyl oleate,
triolein, fatty alcohols and fatty acids methyl esters standard compounds were purchased from SigmaAldrich (Milan,
Italy).
The reagents 14% BF3 in MeOH and hexamethyldisilazane
(HMDS)trimethychlorosilane (TMCS)pyridine (3:1:9) were
purchased from Supelco (Bellefonte, USA). Desferal (deferoxamine methanesulfonate) was purchased from CIBA-Geigy
(Basel, Switzerland). All the other chemicals used in this study
were of analytical grade.
2.3. Lipid extraction and preparation of cholesterol, fatty
alcohols and fatty acids
Lipids were extracted from grated mullet bottarga by the
Folch et al. (1957) procedure. An aliquot of total lipid extract
recovered from the lower chloroform phase was analyzed by
13 C NMR.

The total lipids were quantified by the method of Chiang et

al. (1957).
Separation of cholesterol, fatty alcohols and fatty acids was
obtained by mild saponification. A 5 mg portion of lipids from
each sample was dissolved in 5 ml of ethanol and 100 l of
Desferal solution (25 mg/ml of H2 O), 1 ml of a water solution
of ascorbic acid (25%, w/v), and 0.5 ml of 10N KOH were added.
The mixtures were left in the dark at room temperature for 14 h.
After addition of 10 ml of n-hexane and 7 ml of H2 O, samples
were centrifuged for 1 h at 900 g (Rosa et al., 2005).
The hexane extract containing the unsaponifiable fraction
(cholesterol and fatty alcohols) was collected and the solvent was
evaporated; an aliquot was stored at 20 C until NMR analysis. A portion of the residue was dissolved in 1 ml of MeOH and
injected into the HPLC system. Aliquot of dried fatty alcohols
was converted to trimethylsilyl ether by a mixture of TMCS,
HMDS, and anhydrated pyridine (1:3:9, v/v/v) (200 l) for 2 h
at room temperature, and trimethylsilyl ethers were applied to
capillary gas chromatography.
After addition of further 10 ml of n-hexane to the mixtures,
samples were acidified with 37% HCl to pH 34 and then centrifuged for 1 h at 900 g. The hexane extract with fatty acids,
i.e. the saponifiable phase, was collected and after solvent evaporation an aliquot was stored at 20 C until NMR analysis. A
portion of the dried residue was dissolved in 1 ml of CH3 CN
with 0.14% (v/v) CH3 COOH and aliquots of the samples were
injected into the HPLC system. Aliquot of dried fatty acids was
methylated with 1 ml of 14% BF3 in MeOH (Christie, 1993)
for 30 min at room temperature. After addition of 4 ml of nhexane and 2 ml of H2 O, samples were centrifuged for 20 min
at 900 g. The hexane phase with fatty acids methyl esters was
collected, the solvent was evaporated, the residue was dissolved
in 250 l of n-hexane and aliquots of the samples were injected
into the GC system.
The recovery of fatty acids, fatty alcohols, and cholesterol
was calculated by using an external standard mixture. All solvents evaporation was performed under vacuum.
2.4. HPLC analyses
Analyses of cholesterol and unsaturated fatty acids were
carried out with a Agilent Technologies 1100 liquid chromatograph (Agilent Technologies, Palo Alto, USA) equipped with a
diode array detector. Cholesterol, detected at 203 nm, was measured using a Chrompack column (Chrompack, Middelburg, The
Netherlands), Inertsil 5 ODS-3, 150 mm 3 mm, and MeOH as
mobile phase, at a flow rate of 0.4 ml/min.
Analyses of unsaturated fatty acids and conjugated dienes
fatty acids hydroperoxides (HP), detected at 200 and 234 nm,
respectively, were carried out with a Chrompack column,
Inertsil 5 ODS-2, 150 mm 4.6 mm with a mobile phase of
CH3 CN/H2 O/CH3 COOH (70/30/0.12, v/v/v) at a flow rate of
1.5 ml/min.
The identification of cholesterol, fatty acids, and HP was
made using standard compounds and second derivative as well
as conventional UV spectra, generated using the Agilent Chemstation A.10.02 software (Rosa et al., 2005).

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

2.5. GC analyses
Fatty acids methyl esters and trimethylsilyl derivatives
of fatty alcohols were measured on a gas chromatograph Hewlett-Packard HP-6890 (Hewlett-Packard, Palo Alto,
USA) with flame ionization detector and equipped with
a cyanopropyl methyl-polysiloxane HP-23 FAME column
(30 m 0.32 mm 0.25 m) (Hewlett-Packard). Nitrogen was
used as carrier gas at a flow 2 ml/min. The oven temperature was
set at 175 C, injector temperature 250 C, and detector temperature 300 C. The fatty acids methyl esters and trimethylsilyl
derivatives of fatty alcohols were identified by comparing the
retention times with those of standard compounds. The percentage composition of individual fatty acids and alcohols were

71

calculated using a calibration curve with components injected

at different concentrations, using the Hewlett-Packard A.05.02
software.
2.6. NMR analysis
The extracted lipids (300 mg), the dried saponifiable and
unsaponifiable fractions (100 mg each) were collected and separately dissolved in 0.6 ml of CDCl3 . The samples were placed in
5 mm NMR tubes, stored at 20 C, and analyzed within 2 days.
NMR experiments were performed at 25 C on a Varian
VXR-300 and on a Varian UNITY INOVA 400 spectrometers, operating at the frequency of 75.42 and 100.56 MHz for
carbon, respectively. NOE-suppressed, proton-decoupled 13 C

Fig. 1. The 75 MHz 13 C NMR spectra (1800 ppm) with interpretation of the more relevant resonances of (a) bottarga lipid extract, with enlargement of the spectral
regions between 178.0 and 176.5, 75.0 and 52.5 ppm; (b) saponifiable and (c) unsaponifiable fractions of the extract. All samples are diluted in CDCl3 .

72

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

NMR spectra were recorded. The free induction decay of each

spectrum was acquired with a 2 s acquisition time, a sweep width
of 26 kHz, a 45 pulse angle and a 20 s relaxation delay, 10240
scans were collected. Zero filling and a line broadening of 0.3 Hz
were applied prior to Fourier transformation to minimize the
noise, but not at expense of resolution.
13 C spectra of reference standard compounds were also
obtained using CDCl3 as solvent. All chemical shifts, cited in
ppm, were referred to tetramethylsilane (TMS).
Compositional data of the two products, reported as mol% of
the total lipid extracted, were expressed as means and standard
deviations over six samples of each bottarga product.
3. Results and discussion
3.1.

13 C

NMR analysis of the extracted lipids

A representative 13 C NMR spectrum of the lipid extracts from

bottarga is reported in Fig. 1a. Peak assignments were performed
with the aid of literature data (Gunstone, 1993; Pollesello et
al., 1996; Siddiqui et al., 2003; Scano et al., 2006; Christie,

2007) and by recording spectra of standard compounds. In some

cases, validation of the peak attribution was achieved by adding
standard compounds to the lipid solution and re-recording the
NMR spectrum under the same conditions. The most significant
assignments for each class of compounds are described below.
3.1.1. Carbonyl region (190160 ppm)
In this spectral region the peaks in the range 180177 ppm
were attributed to carboxyl carbons of FFA (docosahexaenoic
acid DHA, eicosapentaenoic acid EPA, and other fatty acids,
see inset Fig. 1a). The intense group of signals centred at 173.84
belongs to the carbonyl carbons in WE, while the remaining
peaks between 173.7 and 172.0 ppm were assigned to carbonyl
carbons of PL, TAG, and CE.
3.1.2. Olenic region (140120 ppm)
In Fig. 2a an expanse of the olefinic region between 132 and
126 ppm is reported. Here, the intense signals at 131.97 and
126.97 ppm were attributed to the -3 and -4 carbon atoms
in n-3 PUFA (in their acid or alcoholic form), respectively. The
remaining olefinic carbons of n-3 PUFA resonate between 129.5

Fig. 2. Olefinic region (132.5126.5 ppm), with interpretation of the more relevant resonances, of the 75 MHz
saponifiable and (c) unsaponifiable fractions of the extract. All samples are diluted in CDCl3 .

13 C

NMR spectra of (a) bottarga lipid extract, (b)

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

73

Fig. 3. 13 C NMR spectrum of bottarga lipid extract in CDCl3 after the addition of DHA (22:6 n-3). Enhanced peaks with respect to the spectrum of the original lipid
matrix, indicated by numbers, are assigned to the following functional groups of DHA: peak 1 belongs to the carboxyl carbon, peaks 29 to olefinic carbons and in
particular peaks 3 and 8 to C4 and C5, respectively. Peak 10 was assigned to C2, peak 11 to CH CH2 CH , peak 12 belongs to C3, peak 13 and peak 14 to -2
and -1 carbons, respectively.

and 127.3 ppm. In this spectral region, signals from the free
and the esterified form of DHA and EPA were identified by
spiking the sample with authentic standard compounds (as an
example the 13 C NMR spectrum of a sample of extracted oil
with the addition of DHA is reported in Fig. 3). In particular, the
resonances at 129.16 and 127.67 ppm were assigned to C4 and
C5 of free DHA respectively (Aursand and Grasdalen, 1992),
while the signal at 129.38 ppm corresponds to C5 of the esterified
DHA. The signals at 128.95 and 128.74 ppm were attributed to
C5 and C6 of free EPA and the peaks at 128.89 and 128.79 ppm
were ascribed to esterified EPA.
The two peaks at 130.42 and 127.50 ppm were ascribed to
the -6 and -7 olefinic carbons of arachidonic acid (AA) and
of other long chain n-6 PUFA, the low intensity of these signals
indicating a small amount of n-6 PUFA in the lipid mixture.
In the region between 130.0 and 129.5 ppm, characteristic
of monounsaturated fatty acids (MUFA), the two main groups
of signals centred at 129.94 and 129.69 ppm were ascribed to
C9 and C10 of the 9 monoenes, respectively, and the peaks at
129.87 and 129.79 ppm were assigned to C11 and C12 of the
11 (Scano et al., 1999).
At lower fields with respect to the resonances ascribable
to the olefinic carbons of fatty acids and alcohols, two small
peaks at 140.68 and 139.54 ppm were observed (Fig. 1a)
and assigned to the C6 involved in the double bond of the
cyclopentaphenanthren-3-ol ring of cholesterol in its free and
esterified form, respectively. In parallel, the high field resonances at 122.47 and 121.48 ppm were ascribed to the C5 olefinic
carbons of CE and Cho, respectively (Pollesello et al., 1996;
Falch et al., 2007). The chemical shift differences of C5 and C6 in
CE compared to Cho are due the different chemical environment
induced by the nearby ester bond in C1 position.
3.1.3. N CH , O CH region (7553 ppm)
In this spectral region the signals from aliphatic carbons
directly bound to an oxygen or to a quaternary nitrogen appear.
These resonances are representative of all the main classes
of fatty compounds and were then used for quantitative purpose. The most prominent cluster of peaks centred at 64.46 ppm

belongs to the O CH2 groups of the alcoholic moiety of

WE. The peaks at 68.80 and 61.98 ppm were attributed to the
O CH and O CH2 functional groups of TAG, respectively, while the N (CH3 )3 of the choline head group in
phosphatidylcholine (PC) resonates at 54.38 ppm. Finally, the
small peaks at 71.56 and 73.57 ppm were ascribed to the C1
carbon atom of Cho and CE, respectively.
3.1.4. Up-eld aliphatic region (3511 ppm)
In this spectral region the peak at 28.57 ppm was attributed
to the O CH2 CH2 group of WE and the well-resolved signal
at 20.45 ppm to the -2 methylene group of all the n-3 fatty
acids and alcohols. The signal at 11.79 ppm was ascribed to the
CH3 group (C18) of Cho and CE. No signal characteristic of
trans double bonds in fatty acids at 30 and 32 ppm (Christie,
2007) was found, according to previous data obtained by infrared
spectroscopy (Iyengar and Schlenk, 1967).
3.2. Semi-quantitative NMR data
Table 1 shows the composition of the major lipid classes
estimated by integration of suitable 13 C NMR resonances of
the extracted oil. All samples showed similar contents of TAG,
Table 1
Lipid class composition of bottarga products calculated from the integrated areas
of 13 C NMR spectra
Percentage of total lipids (mol%)a
Product 1
WE
TAG
FFA
Cho
CE
PC

63.6
6.7
19.4
2.7
1.5
6.1

2.7
0.8
3.9
0.4
0.7
1.1

Product 2
51.2
8.7
28.8
3.6
1.9
5.8

0.7
0.6
0.5
0.1
1.2
0.2

Relative contents were measured from normalized areas of the following peaks:
64.46 ppm for WE, 61.98 ppm for TAG, range 180177 ppm for FFA, 71.56 ppm
for Cho, 73.57 ppm for CE, 54.38 ppm for PC.
a Mean and standard deviation over six samples.

74

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

Cho, CE and PC, while different quantity of FFA and WE were

detected in the samples of product 1 with respect to those of
product 2.
The high FFA content (means of 19.4 and 28.8 mol%, for
products 1 and 2, respectively) found in all samples is to be
ascribed to hydrolysis processes on the esterified molecules
of the original lipid matrix during manufacturing and storage,
according to previous investigations on fish roes (Basby et al.,
1998; Falch et al., 2007). Therefore, due to the high WE content of bottarga lipids, in addition to the release of FFA from the
original lipid matrix also the free fatty alcohol counterparts from
the hydrolysis processes are to be expected. However, due to
the marked overlapping, the typical resonances of free alcoholic
derivatives were not identified in the 13 C NMR spectra, but in the
corresponding 1 H NMR spectra (spectra not shown) at 3.62 ppm
a triplet signal appears, attributed to the HO CH2 CH2 group
of free fatty alcohols. On the other hand, NMR signals originating from residual fractions of TAG and PC after their hydrolysis
were below detection.
3.3. 13 C NMR spectra of the saponiable and
unsaponiable fractions
Further information on the molecular composition of the lipid
extracts of bottarga were gained from the analysis of the 13 C
NMR spectra of both the saponifiable (Figs. 1b and 2b) and the
unsaponifiable (Figs. 1c and 2c) fractions. As expected, in the
former spectrum fatty acids from hydrolysis of WE, TAG, PL,
and CE, together with the original FFA were detected, while in
the latter spectrum the alcoholic lipid derivatives, i.e. Cho and
the alcoholic moieties of WE, were found.
From the comparison of these spectra with that of the pure
extracted lipids (Figs. 1a and 2a) the following considerations
can be drawn:
The carbonyl region of the saponifiable matter exhibited an
enhancement of resonances from FFA (180177 ppm), while,
as expected, no carbonyl peaks were observed in the spectrum
of the unsaponifiable fraction.
In the olefinic region of the unsaponifiable fraction only
the typical resonances (spectral range between 130.0 and
129.5 ppm, Fig. 2c) of monoenes were found. The saponifiable fraction showed, in addition to MUFA, also a number of
resonances ascribed to PUFA, among which DHA and EPA,
were clearly detectable. Furthermore, the unsaponifiable fraction lacked in the olefinic signals of CE detected in Fig. 1a,
while the characteristic resonances at 140.71 and 121.65 ppm
of C6 and C5 in Cho were enhanced.
In the unsaponifiable fraction the intense peak at 63.09 ppm
(Fig. 1c) was ascribed to the HO CH2 functional group
of fatty alcohols, exhibiting an up-field shift with respect to
the corresponding peak of the esterified form (64.46 ppm in
Fig. 1a); in parallel, the signal at 32.80 ppm, attributed to
the HO CH2 CH2 group, resulted down-field shifted as
compared to the corresponding esterified form at 28.57 ppm.
These attributions are in agreement with previous studies
(Gunstone, 1993).

Table 2
Fatty alcohols composition (%) of bottarga products by GC
Fatty alcohol

Product 1

14:0
15:0
16:0
18:0
16:1 n-7
18:1 n-7
18:1 n-9

10.9
3.1
52.8
6.8
10.4
3.3
3.9

SFA
MUFA

73.6 0.5
17.7 0.2

0.1
0.0
0.4
0.1
0.1
0.1
0.1

Product 2
7.3
3.3
51.7
7.9
9.4
3.6
4.4

0.1
0.2
0.6
0.1
0.3
0.1
0.1

70.2 0.5
17.3 0.4

SFA, saturated fatty alcohols; MUFA, monounsaturated fatty alcohols. Mean

and standard deviation over six samples.

3.4. Fatty alcohols and fatty acids analyses by GC

Quali-quantitative information on the individual fatty acids
and fatty alcohols that compose the major lipid classes of bottarga were obtained by GC analysis. The analyzed samples
exhibited similar fatty acids and alcohols profiles.
The content of fatty alcohols present in the unsaponifiable
matter is reported in Table 2 as percentage of the total amount
of alcohols. There was no considerable variation between the
analyzed samples. As found in previous investigations (Iyengar
and Schlenk, 1967; Bernasconi et al., 2007) only saturated (C14
to C18 chain-length) and monounsaturated (16:1 and 18:1 isomers) alcohol derivatives were found, in particular, the samples
were characterized by a high amount of 16:0 (about 52%).
The composition of fatty acids present in the saponifiable
matter, that comprise the original FFA and those derived from
induced hydrolysis of the intact molecules, of the two bottarga
products are shown in Table 3 and expressed as percentage of
total fatty acids. They showed a concentration of approximately
1618% of saturated fatty acids (mainly 16:0), 35% of MUFA
(mainly 16:1 n-7, and 18:1 n-9), and 34-37% of PUFA. In particular, the total content of the n-3 derivatives EPA and DHA
amounted to ca. 22%. Low concentrations of n-6 fatty acids
(45%) were found, thus resulting in an elevate (n-3)/(n-6) ratio,
in agreement with previous works on fish roes (Wiegand, 1996;
Falch et al., 2006). Finally, the very low contents of 20:1 and
longer monounsaturated fatty acids and alcohols found in the
samples suggest that the strong selection against the incorporation of these monoenes in fish eggs (Wiegand, 1996) exists also
in mullet roes.
3.5. Main lipid characteristics
An average lipid content of 270 mg/g of the edible portion
for bottarga samples was measured. By HPLC, the total cholesterol was measured as mean content of 9.3 and 10.5 mg/g of the
edible portion for samples of products 1 and 2, these values represented ca. 34% of total lipids, that is, a concentration lower
than in chicken egg (Yannakopoulos et al., 2005). Furthermore,
the highly unsaturated fatty acids n-3 content in the edible portion was detected as follows: 20.5 and 23.2 mg/g of DHA, 13.5
and 9.9 mg/g of EPA, and a minor amount of 22:5 n-3 (9.7 and

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

Table 3
Fatty acids composition (%) of bottarga products by GC
Fatty acid

Product 1

Product 2

12:0
14:0
15:0
16:0
18:0
20:0
16:1 n-7
18:1 n-7
18:1 n-9
20:1 n-9
16:2
16:3
16:4
18:2 n-6
18:3 n-3
18:3 n-6
18:4 n-3
20:3 n-3
20:3 n-6
20:3 n-9
20:4 n-6
20:5 n-3
22:4 n-6
22:5 n-3
22:6 n-3

0.03 0.01
2.07 0.11
0.44 0.08
10.90 0.19
3.18 0.12
0.15 0.07
17.92 0.43
7.22 0.20
10.61 0.13
0.16 0.02
0.70 0.03
1.46 0.14
1.98 0.45
1.27 0.04
0.57 0.02
0.40 0.01
1.42 0.03
Trace
0.25 0.05
0.11 0.05
1.69 0.08
9.40 0.17
0.49 0.06
5.95 0.12
11.62 0.20

0.04 0.01
1.75 0.06
0.52 0.02
12.69 0.28
3.07 0.07
0.14 0.08
14.71 0.35
6.23 0.11
13.70 0.18
0.37 0.04
0.56 0.02
0.63 0.18
Trace
1.58 0.14
0.97 0.05
0.38 0.08
1.16 0.05
0.18 0.03
0.15 0.00
0.29 0.15
2.21 0.10
6.73 0.31
0.48 0.14
3.96 0.14
15.28 1.03

SFA
MUFA
PUFA

16.77 0.27
35.90 0.32
37.27 0.36

18.22 0.43
35.01 0.27
34.47 1.48

SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids. Mean and standard deviation over six samples.

by HPLC as HP, was comparable to that of fresh marine oils

(Alamed et al., 2006).
To note, FFA can originate from hydrolysis processes on the
original lipid matrix and previous studies on fresh (Falch et al.,
2007) and salted (Basby et al., 1998) fish eggs demonstrated that
the extent of this deterioration process is strongly influenced by
storage conditions. This finding can represent a serious drawback when either the fresh roe or the final bottarga products are
submitted to long-term storage.
Due to the high content of WE in lipid mixture valuable information were also given by the analysis of the 13 C NMR spectra of
both the saponifiable and unsaponifiable matters, and the comparison with the more detailed HPLC and GC data allowed a
complete characterization of the fatty alcohols and fatty acids
that compose the WE, TAG, PL, and CE.
The two commercial products here examined show similar
composition of fatty acids and alcohols as calculated by GC
technique. Semi-quantitative NMR data indicate a different content of WE and FFA, this result suggests that probably product
2 underwent major hydrolysis process compared to product 1.
The bottarga samples were found to be rich in n-3 PUFA, as
DHA and EPA, that have been recognised as having an important role in health (Ruxton et al., 2004). Considering the high
content of WE in bottarga, we can suppose that a significant
amount of these n-3 fatty acids are WE components, as previously observed (Bernasconi et al., 2007). In the light of a
recent study that demonstrated that WE enriched in n-3 fatty
acids have a low degree of susceptibility to oxidation (Gorreta
et al., 2002), bottarga may be regarded as stable natural source
of health beneficial n-3 fatty acids.

5.5 mg/g), as mean values over six samples for products 1 and
2, respectively.
The oxidation status of fatty acids was also measured by
HPLC detection of HP. The level of HP in the bottarga samples
was ca. 0.1 mol/g of the edible portion (0.4 nmol/mg of lipid).

Acknowledgments

4. Conclusions

References

In the present work, by the conjunct use of NMR and chromatographic techniques, we have exhaustively studied the major
intact lipid classes and the individual fatty acids and alcohols
of the lipid extracted from commercial products of bottarga
manufactured in Sardinia.
We have assigned the most important resonances in the 13 C
NMR spectrum of this complex lipid mixture and the major lipid
classes were measured as mol%. The more represented lipid
components were WE. In agreement with previous studies on
fish eggs (Bledsoe et al., 2003; Falch et al., 2006) among the PL
classes, only PC was found. Besides cholesterol, also cholesterol
esters, probably formed during storage of the fresh eggs (Falch et
al., 2007), were detected in the 13 C NMR spectra. Furthermore,
a relatively high quantity of FFA was reported, and the presence
of DHA and EPA in their free form can be inferred from the
occurrence of specific resonances in the carbonyl, olefinic and
aliphatic spectral regions of the 13 C NMR spectra. In spite of the
high content of FFA the level of oxidative products, measured

75

We thank Nicoletta Zinnarosu for her expert NMR technical assistance, and Stefano Rocca for helpful discussions on
manufacturing mullet roes.

Alamed, J., McClements, D.J., Decker, E.A., 2006. Influence of heat processing
and calcium ions on the ability of EDTA to inhibit lipid oxidation in oil-inwater emulsions containing omega-3 fatty acids. Food Chem. 95, 585590.
Aursand, M., Grasdalen, H., 1992. Interpretation of the 13 C spectra of omega-3
fatty acids and lipid extracted from white muscle of Atlantic Salmon (Salmo
salar). Chem. Phys. Lipids 62, 239251.
Aursand, M., Rainuzzo, J.R., Grasdalen, H., 1993. Quantitative high-resolution
13 C and 1 H nuclear magnetic resonance of -3 fatty acids from white muscle
of Atlantic Salmon (Salmo salar). JAOCS 70, 971981.
Aursand, M., Standal, I.B., Axelson, D.E., 2007. High-resolution 13 C nuclear
magnetic resonance spectroscopy pattern recognition of fish oil capsules. J.
Agric. Food Chem. 55, 3847.
Basby, M., Jeppesen, V.F., Huss, H.H., 1998. Chemical composition of fresh and
salted Lumpfish (Cyclopterus lumpus) roe. J. Aquat. Food Prod. Technol. 7,
721.
Bernasconi, R., Bolzacchini, E., Galliani, G., Gugliersi, F., Rindone, B.,
Rindone, M., Tacconi, M.T., Terraneo, A., 2007. Determination of the content of wax esters in some sea food and their molecular composition. A
comparison with -3 enriched wax esters. Lebensm-Wiss. Technol. 40,
569573.
Bledsoe, G.E., Bledsoe, C.D., Rasco, B., 2003. Caviars and fish roe products.
Crit. Rev. Food Sci. Nutr. 43, 317356.

76

P. Scano et al. / Chemistry and Physics of Lipids 151 (2008) 6976

Casu, M., Anderson, G.J., Choi, G., Gibbons, W.A., 1991. NMR lipid profiles
of cells, tissues and body fluids-1D and 2D proton NMR of lipids from rat
liver. Magn. Res. Chem. 29, 594602.
Chiang, S.D., Gessert, C.F., Lowry, O.H., 1957. Colorimetric determination of
extracted lipids. An adaptation for microgram amounts of lipids obtained
from cerumen. Curr. List. Med. Lit. Res. Rep. 33, 5664.
Christie, W.W., 1993. Preparation of ester derivatives of fatty acids for chromatographic analysis. In: Advantage in Lipid MethodologyTwo. The Oily
Press, Dundee, Scotland, pp. 69111.
Christie, W.W., The Lipid Library, www.lipidlibrary.co.uk and the literature
therein cited. Accessed April 2007.
Falch, E., Storseth, T.R., Aursand, M., 2006. Multi-component analysis of
marine lipids in fish gonads with emphasis on phospholipids using high
resolution NMR spectroscopy. Chem. Phys. Lipids 144, 416.
Falch, E., Storseth, T.R., Aursand, M., 2007. High resolution NMR for studying
lipid hydrolysis and esterification in cod (Gadus morhua) gonads. Chem.
Phys. Lipid. 147, 4657.
Folch, J., Lees, M., Sloane-Stanley, G.H., 1957. A simple method for the isolation and purification of total lipid from animals tissues. J. Biol. Chem. 226,
497509.
Gorreta, F., Bernasconi, R., Galliani, G., Salmona, M., Tacconi, M.T., Bianchi,
R., 2002. Wax esters of n-3 polyunsaturated fatty acids: a new stable formulation as a potential food supplement. 1Digestion and absorption in rats.
Lebensm-Wiss. Technol. 35, 458465.
Gribbestad, I.S., Aursand, M., Martinez, I., 2005. High resolution 1 H magnetic
resonance spectroscopy of whole fish, fillets and extracts of farmed atlantic
salmon (Salmo salar) for quality assessment and compositional analyses.
Aquaculture 250, 445457.
Gunstone, F.D., 1993. High-resolution 13 C spectra of long-chain acids, methyl
esters, glycerol esters, wax esters, nitriles, amides, alcohols and acetates.
Chem. Phys. Lipids 66, 189193.
Iyengar, R., Schlenk, H., 1967. Wax esters in mullet (Mugil cephalus) roe oil.
Biochem. 6, 396402.
Joh, Y.G., Brechany, E.Y., Christie, W.W., 1995. Characterization of wax esters
in the roe oil of amber fish, Seriola aureovittata. J. Am. Oil Chem. Soc. 22,
707713.

Lu, J.Y., Ma, Y.M., Williams, C., Chung, R.A., 1979. Fatty and amino acid
composition of salted mullet roe. J. Food Sci. 44, 676677.
Nevenzel, J.C., 1969. Occurrence, function and biosynthesis of wax esters in
marine organisms. Lipids 5, 308319.
Pollesello, P., Eriksson, O., Hockerstedt, K., 1996. Analysis of total lipid extracts
from human liver by 13 C and 1 H nuclear magnetic resonance spectroscopy.
Anal. Biochem. 236, 4148.
Rosa, A., Deiana, M., Corona, G., Atzeni, A., Incani, A., Dess`, M.A., 2005.
Cholesterol as target of Fe-NTA induced peroxidation in rats tissue. Toxicol.
Lett. 157 (1), 18.
Ruxton, C.H.S., Reed, S.C., Simpson, M.J.A., Millington, K.J., 2004. The health
benefits of omega-3 polyunsaturated fatty acids: a review of the evidence. J.
Hum. Nutr. Diet. 17, 449459.
Sacchi, R., Medina, I., Aubourg, S.P., Giudicianni, I., Paolillo, L., Addeo, F.,
1993. Quantitative high-resolution 13 C-NMR analysis of lipids extracted
from the white Muscle of Atlantic Tuna (Thunnus alalunga). J. Agric. Food
Chem. 41, 12471253.
Scano, P., Casu, M., Lai, A., Saba, G., Dess`, M.A., Deiana, M., Corongiu,
F.P., Bandino, G., 1999. Recognition and quantitation of cis-vaccenic and
eicosenoic fatty acids in olive oils by 13 C nuclear magnetic resonance spectroscopy. Lipids 34, 757759.
Scano, P., Maxia, C., Maggiani, F., Crnjar, R., Lai, A., Sirigu, P., 2005. A
histological and NMR study of the melon of the striped dolphin (Stenella
coeruleoalba). Chem. Phys. Lipids 134, 2128.
Scano, P., Cesare Marincola, F., Locci, E., Lai, A., 2006. 1 H and 13 C NMR studies
of melon and head blubber of the striped Dolphin (Stenella coeruleoalba).
Lipids 41, 10391048.
Siddiqui, N., Sim, J., Silwood, C.J.L., Toms, H., Iles, R.A., Grootveld, M.,
2003. Multicomponent analysis of encapsulated marine oil supplements
using high-resolution 1 H and 13 C NMR techniques. J. Lipid Res. 44, 2406
2427.
Wiegand, M.D., 1996. Composition, accumulation and utilization of yolk lipids
in teleost fish. Rev. Fish Biol. Fisheries 6, 259286.
Yannakopoulos, A., Tserveni-Gousi, A., Christaki, E., 2005. Enhanced egg production in practice: the case of Bio-Omega-3 Egg. Int. J. Poult. Sci. 4,
531535.