Process development of Lappula squarrosa oil refinement: Monitoring of pyrrolizidine alkaloids in boraginaceae seed oils

Article abstract of DOI:10.1007/s11746-014-2426-1

Boraginaceous plants are characterized by high levels of polyunsaturated fatty acids and show a high ratio of ω-3/ω-6 fatty acids. In addition, Lappula squarrosa (Boraginaceae) shows high levels of stearidonic acid content (ω-3; 18:4; 6,9,12,15-octadecatetraenoic acid) showing interesting medical and health promoting properties. On the other hand Boraginaceous plants contain genotoxic and carcinogenic pyrrolizidine alkaloids (PA). An HPLC-ESI-MS/MS sum parameter method was developed to monitor the total sum of 1,2-unsaturated PA in seed oil. The method was used to monitor different steps in oil refinement using lab model experiments and pilot scale refinement of L. squarrosa seed oil. A limit of detection and a limit of quantification (LOQ) of 0.02 μg retronecine equivalents (RE)/kg and 0.05 μg RE/kg were achieved, respectively. Multiple washing steps at pH 2.2 can significantly reduce the PA content to 0.07 % of the start value. In addition, combining washing with neutralization, bleaching and deodorization can reduce the PA content of L. squarrosa oil below the LOQ (0.05 μg RE/kg). The newly established method was further used to analyze the PA content of commercially available Boraginaceous seed oils (Echium spp., Borago officinalis). Three out of ten products were tested as PA positive (PA content ranging from not detected to 0.6 μg RE/kg product).

Full text of DOI:10.1007/s11746-014-2426-1

J Am Oil Chem Soc
DOI 10.1007/s11746-014-2426-1
ORIGINAL PAPER
Process Development of Lappula squarrosa Oil Refinement:
Monitoring of Pyrrolizidine Alkaloids in Boraginaceae Seed Oils
•
•
•
Luise Cramer Gunther Fleck Gert Horn
Till Beuerle
Received: 12 September 2013 / Revised: 4 January 2014 / Accepted: 23 January 2014
Ó AOCS 2014
Abstract Boraginaceous plants are characterized by high
levels of polyunsaturated fatty acids and show a high ratio
of x-3/x-6 fatty acids. In addition, Lappula squarrosa
(Boraginaceae) shows high levels of stearidonic acid con-
tent (x-3; 18:4; 6,9,12,15-octadecatetraenoic acid) showing
interesting medical and health promoting properties. On the
other hand Boraginaceous plants contain genotoxic and
carcinogenic pyrrolizidine alkaloids (PA). An HPLC–ESI–
MS/MS sum parameter method was developed to monitor
the total sum of 1,2-unsaturated PA in seed oil. The method
was used to monitor different steps in oil refinement using
lab model experiments and pilot scale refinement of L.
squarrosa seed oil. A limit of detection and a limit of
quantification (LOQ) of 0.02 lg retronecine equivalents
(RE)/kg and 0.05 lg RE/kg were achieved, respectively.
Multiple washing steps at pH 2.2 can significantly reduce
the PA content to 0.07 % of the start value. In addition,
combining washing with neutralization, bleaching and
deodorization can reduce the PA content of L. squarrosa
oil below the LOQ (0.05 lg RE/kg). The newly established
method was further used to analyze the PA content of
commercially available Boraginaceous seed oils (Echium
spp., Borago officinalis). Three out of ten products were
tested as PA positive (PA content ranging from not
detected to 0.6 lg RE/kg product).
Keywords Pyrrolizidine alkaloids Á Lappula squarrosa Á
Oil refinement Á Boraginaceous seed oils Á HPLC–ESI–MS/
MS Á Food supplements Á Borago officinalis Á Echium spp.
Abbreviations
BMDL10 Benchmark dose lower confidence limit is the
95 % lower confidence limit of the benchmark
dose associated with a 10 % response
ND
Not detected
Introduction
The polyunsaturated fatty acids (PUFA) a-linolenic acid
(ALA; x-3; 18:3; 9,12,15-octadecatrienoic acid) and lino-
leic acid (LA; x-6; 18:2; 9,12-octadecadienoic acid) are
known to be essential for humans. Starting from ALA (x-
3) or LA (x-6) humans/animals convert these two precur-
sors into the long chain PUFA sharing the same set of
enzymes (desaturases and elongases). This results in two
series (x-3/x-6) of long chain PUFA like eicosapentaenoic
acid (EPA; x-3; 20:5; 5,8,11,14,17-eicosapentaenoic acid),
docosapentaenoic acid (DPA; x-3; 22:5; 7,10,13,16,19-
docosapentaenoic acid), docosahexaenoic acid (DHA; x-3;
22:6; 4,7,10,13,16,19-docosahexaenoic acid) or arachi-
donic acid (AA; x-6; 20:4; 5,8,11,14-eicosatetraenoic
acid), respectively [1]. While long chain PUFA play an
essential role as structural components of cell membranes
the C20-PUFA, EPA (x-3) and AA (x-6) can be trans-
formed into eicosanoids including prostaglandins, prosta-
cyclins, thromboxanes and leukotrienes, which show a
multitude of health and disease modulating properties [2,
L. Cramer Á T. Beuerle (&)
¨
Institut fu¨r Pharmazeutische Biologie, Technische Universitat
Braunschweig, Mendelssohnstr. 1, 38106 Braunschweig,
Germany
e-mail: t.beuerle@tu-bs.de
G. Fleck
Pilot Pflanzenoeltechnologie Magdeburg e. V. (PPM),
Magdeburg, Germany
G. Horn
Exsemine GmbH, Zappendorf, Germany
123

J Am Oil Chem Soc
3]. Plant oils do not contain those long chain x-3/x-6
PUFA and vary considerably in the content of saturated/
unsaturated fatty acids as well as in the ratio of x-3/x-6
fatty acids. While the European Food Safety Authority
(EFSA) does not recommend a specific ratio of x-3/x-6
fatty acids in its opinion on dietary reference values [1],
some national authorities like the Deutsche Gesellschaft fu¨r
Lappula is a genus of about 50 species within the
Boraginaceae family commonly found in temperate Eur-
asian regions. Lappula squarrosa is an annual or biennial
plant and prefers warmth, dry and sandy soils. Its natural
occurrence is rare and scattered through Central Europe.
The common name European stickseed relates to the fruits
which are clusters of four nutlets which are coated in
hooked prickles helping in the dispersal by affixing them-
selves to animal coats etc. [16]. The seeds are known to
contain *17–19 % SDA [17–20]. On the other hand, L.
squarrosa is also a PA-containing plant which needs to be
considered when the oil is intended for human
consumption.
¨
Ernahrung suggest a x-3/x-6 of 1:5 or higher as beneficial
[4]. Based on Food Consumption Studies in Germany
(years 1994–1998) the actual x-3/x-6 ratio is instead
1:7.2–8.6 in western diets [4].
Seed oils from Boraginaceous plants hold a special
position concerning their fatty acid composition. They are
characterized by high levels of PUFA and at the same time
a high ratio x-3/x-6 compared to the commonly used
edible oils like olive, canola or palm oil (low in PUFA) or
safflower, sunflower or soybean oils (rich in PUFA but high
in x-6 LA) [5]. In addition, these Boraginaceous plants
show high levels of stearidonic acid (SDA; x-3; 18:4;
6,9,12,15-octadecatetraenoic acid). SDA is only found in
few higher plant oils and has drawn some interests due to
its medicinal and health promoting properties [6–9].
In humans the conversion of ALA, catalyzed by the
delta-6 desaturase, is limited. Oils rich in SDA can be more
efficiently converted into longer chain PUFA like EPA,
DPA, or DHA [9]. Because of this, SDA rich plant seed oils
are considered as valuable sources to bypass recommended
EPA/DHA (x-3) dietary intakes without further increasing
overexploitation of the aquatic environment. Although
transgenic varieties of soybean were developed that pro-
duce up to 30 % SDA the negative perception of transgenic
products may limit their economic potential [9, 10]. Hence,
natural occurring SDA-rich plants are a promising and
sustainable source for the production of SDA-rich oils.
So far, approval has been given to refined Echium
plantagineum seed oil as novel food ingredient under
Regulation (EC) No. 258/97 of the European Parliament
and of the Council. The SDA content of this oil is specified
as ‘‘not \10 %’’ [11]. Typically, E. plantagineum seed oil
shows a SDA content range of 9.6–15.5 % [12, 13].
On the other hand, Boraginaceous plants are known to
contain toxic 1,2-unsaturated pyrrolizidine alkaloids (PA)
[14]. PA are toxic secondary plant compounds which are
constitutively expressed as a chemical defense strategy in
some plant families [14]. Recently, the EFSA has pub-
lished a scientific opinion on PA in food and feed [15].
Based on a 2-year carcinogenicity study of lasiocarpine in
rats, it was concluded that the PA exposure from food
should be as low as possible [15]. Because of the genotoxic
and carcinogenic nature of 1,2-unsaturated PA the approval
of refined E. plantagineum seed oil includes the specifi-
cation of \4 lg PA/kg oil which represents the detection
limit of the TLC method used in this application.
Here, we report the application of an HPLC–ESI–MS/
MS method which was used to monitor PA content at
different stages of oil refinement. The method was finally
used in terms of a preventive consumer protection to
optimize and implement an effective refinement procedure
for L. squarrosa seed oil. In addition, the PA content of
several Boraginaceous plant seed oils which are commer-
cially available on the European market for cooking, food
supplements, or cosmetic purposes were analyzed.
Experimental Procedures
Chemicals
All chemicals used were purchased from Roth (Karlsruhe,
Germany) and Sigma-Aldrich (Seelze, Germany) and were
of HPLC grade purity or redistilled before use. Lithium
aluminium hydride solution AcroSealÒ (1 M in THF) and
pyridine AcroSealÒ were acquired from Acros Organics
(New Jersey, USA). For spiking experiments and method
development, the certified reference material of senecio-
nine (Sen), senecionine-N-oxide (SenNOx) and seneciph-
ylline were purchased from PhytoLab (Vestenbergsgreuth,
Germany).
Sun Flower Oil as a Matrix for Spiking Experiments
Common sunflower oil for cooking and frying was
obtained from a local supermarket. The absence of PA and
the corresponding PA-N-oxides (PANO) was tested and
verified before it was used as a matrix in method devel-
opment to optimize the refinement process.
Lappula squarrosa Plant Cultivation and Oil
Production
L. squarrosa was grown under controlled organic cultiva-
tion conditions on medium to good soils located in
¨
the region ‘‘Ostliches Harzvorland’’ (Saxony-Anhalt,
123

J Am Oil Chem Soc
Germany). Climate conditions: annual mean temperature is
9.0 °C, annual average rainfall below 500 mm, and ele-
vations below 100 m. Cultivation, harvest and seed
cleaning was performed with standard agricultural
equipment.
(Su¨d-Chemie, Mu¨nchen, Germany). The additive was
immediately removed by pressurized filtration using a
cellulose acetate filter (OE 67, 0.45 lm; Whatman Inter-
national Ltd, Springfield Mill, UK) using nitrogen at a
pressure of 3 bar. All steps were performed under an inert
atmosphere using nitrogen gas, while residual oxygen was
monitored throughout the process using an electronic
oxygen sensor device (GOX 100, Greisinger Electronic
GmbH, Regenstauf, Germany). The oxygen levels ranged
from 1.7 to 1.9 %.
Seed oil production was carried out by pressing the
cleaned L. squarrosa seeds using a Monforts Komet S 87
¨
G/1 (IBG Monforts Oekotec GmbH & Co. KG, Mon-
chengladbach, Germany) at 3.0–4.5 U/min, jet 15 mm,
pressing head temperature 45–60 °C using an inert atmo-
sphere of nitrogen. The oxygen levels in the oil collection
container were below 1 % during pressing. The oil content
of the L. squarrosa seeds ranged 19.8–23.7 %, while the oil
content of the residues was 4–5 %.
Pilot Scale Refinement of Lappula squarrosa Seed Oil
Crude L. squarrosa seed oil was filtered using a chamber
filter press (Petkus FP 250-16/10-20 (filter cloth: PPM44/
025/14/660; PETKUS Technologie GmbH, Wutha-Farn-
roda, Germany) and 1 % of Claracel DIC (CECA, La
Garenne-Colombes, France) as filtration additive. The
resulting oil (one batch 50 kg) was further processed in an
Oil Refinement Procedures
Washing and Deodorization Conditions to Reduce PA
Content of Plant Seed Oil
¨
oil refinement mini plant (OHMI Engineering GmbH,
First, 1.5 mg senecionine and 1.5 mg SenNOx (stock
solution in methanol) were added to 13.6 kg of commer-
cially available sunflower oil. The mixture was homoge-
nized for 30 min at 15 °C, yielding a PA spiked starting
material. Two aliquots of 500 ml were put aside to serve as
a reference material for later PA analysis.
Magdeburg, Deutschland) using nitrogen as the inert
atmosphere throughout the whole process. The different
conditions for each step are briefly summarized. One
washing step at pH 2.2 (5 l each); neutralization in two
steps using appropriate amounts of aqueous 7 % NaOH
based on the TAN followed by five washing steps with
water. These steps were conducted at 90 °C and 20 min
processing time for each step. Drying and bleaching
including a final filtration were conducted under a vacuum
(25–30 mbar). One percent Fuller’s earth was added as a
filtration additive. Final deodorization was achieved using
a temperature of 240 °C for 65 min at 3–5 mbar.
At each step in the process development, oil samples
were collected and used for PA analysis.
Washing experiments were conducted at pH values of
2.2, 3.0, 4.0, 5.0, 6.0 and 7.0 using 500 g of the PA spiked
material and 50 ml of the corresponding aqueous washing
solution. The appropriate pH of a citric acid-phosphate
buffer solution was obtained by mixing appropriate vol-
umes of citric acid and disodium phosphate buffers of 100
and 200 mM, respectively. The final pH was adjusted using
a pH-meter (WTW Wissenschaftlich-Technische Wer-
Commercial Boraginaceous Plant Seed Oils
Ten different Boraginaceous plant seed oils and food
supplement capsules containing Boraginaceous plant seed
oils were purchased from several supermarkets, pharmacies
and German language internet shops. The products con-
tained either Echium spp. or Borago officinalis seed oil and
the intended use was food, food supplement and cosmetics.
¨
kstatten GmbH, Weilheim, Germany). Washing was per-
formed in a stirring vessel at 85 °C for 15 min. Phases
were separated after 30 min settling time and the resulting
oil phase as used for PA analysis.
Identical washing experiments were conducted using L.
squarrosa seed oil.
In addition, a lab scale refinement was conducted using
a modified 2-l three-neck sulfuration flask containing a
draining outlet. Following an initial washing step
(1,750 ml oil and 175 ml citric acid-phosphate buffer
solution pH 7.0 or 2.2, respectively), the oil fraction was
neutralized using appropriate amounts of aqueous 7 %
NaOH based on the total acid number (TAN). Drying and
bleaching were performed in a 2-l three-neck sulfuration
flask at 90 °C. Drying was achieved by stirring under
vacuum (30–40 mbar) for 60 min. Additional bleaching
was performed by adding 1 % Tonsil Supreme 114 FF
Sample Preparation Procedures
Sample Preparation of PA Spiking Experiments
in Sunflower Oil
Twenty grams of oil was emulsified with 18 ml 0.05 M
sulfuric acid. One hundred and ten nanograms seneciph-
ylline (11 ll of a 10 lg/ml stock solution in methanol) was
added as the internal standard. The mixture was shaken for
3 min, followed by centrifugation (10 min at 12,0969g at
22 °C). The oil phase was extracted one more time with
123

J Am Oil Chem Soc
5 ml 0.05 M sulfuric acid. The aqueous phases were
combined and 8 ml dichloromethane was added. The
mixture was centrifuged (2 min at 12,0969g at 4 °C) and
the aqueous phase was separated (pipetted off) and
extracted with 8 ml dichloromethane again. Then, 0.5 g
zinc dust was added to convert the PA-N-oxide to the
corresponding tertiary PA. This mixture was stirred for 3 h
at room temperature. Afterwards, the mixture was alkalized
with 4 ml NH4OHconc to a pH value of around 12 and
applied on a 20-g IsoluteÒ-column (Biotage, Eke, Bel-
gium). After an exposure time of 20 min the PA were
eluted with 2 9 40 ml dichloromethane and the organic
phase was dried under a stream of nitrogen gas. After re-
dissolving in 0.5 ml deionized water and 0.5 ml acetoni-
trile the sample was directly analyzed by HPLC–ESI–MS/
MS.
a stream of nitrogen gas and redissolved in 50 ll methanol.
Six hundred microliters LiAlH4 (1 M in THF) were added
and the mixture was sealed, immediately vortexed and kept
at 4 °C for 30 min. After that, the reaction was stopped by
adding 1 ml dichloromethane and 200 ll NaOH (10 %).
0.5 g sodium sulfate was added and the mixture was briefly
vortexed. The organic phase was extracted by filtrating
through glass wool. The extraction was repeated three more
times with 6 ml dichloromethane each time. The resulting
organic phases were dried under a stream of nitrogen. For
derivatization, the residue was redissolved with 0.06 g
phthalic acid in 0.3 ml pyridine and kept at 85 °C for
60 min. After a cool-down-period of 2 min the reaction
was stopped by adding 0.7 ml deionized water. The vor-
texed sample was measured by HPLC–ESI–MS/MS the
same day.
Sample Preparation of PA Spiking Experiments in Lappula
squarrosa Oil
Sample Preparation of Commercial Boraginaceous Plant
Seed Oils
The absence of senecionine, SenNOx and seneciphylline
was confirmed prior to the method development.
The same sample work-up was performed with commercial
Boraginaceous plant seed oils and food supplement cap-
sules as described above (see ‘‘Sample Preparation for Lab
and Pilot Scale Refinement of L. squarrosa Oil’’) In the
case of food supplement capsules the capsules were either
cut open with a scalpel to press out the oil or they were
directly mixed with 0.05 M sulfuric acid and shaken till the
soft capsule burst open.
A similar sample preparation was performed on L.
squarrosa oil as described above for sunflower oil. Dif-
ferences were extraction three times, first with 15 ml then
two times using 7.5 ml of 0.05 M sulfuric acid.
Sample Preparation for Lab and Pilot Scale Refinement
of Lappula squarrosa Oil
Analytical Methods
The sample work-up was based on the recent GC–MS sum
parameter method [22, 23]. Major modifications to sim-
plify and to increase time efficiency were introduced and
are briefly described here.
HPLC–ESI–MS/MS Method for PA Spiking Experiments
in Sunflower Oil and Lappula squarrosa Oil
Two to 20 g L. squarrosa oil (depending on the
expected PA content of the sample) were emulsified with
6–10 ml 0.05 M sulfuric acid. 7-O,9-O-dibutyroyl-
[9,9-2H2]-retronecine was added as an internal standard
(30–100 ng according to the expected PA content) from a
2 lg/ml stock solution in methanol. The mixture was
shaken for 3 min and the two phases were separated by
centrifugation (10 min at 12,0969g at 22 °C). The
extraction was repeated two more times with 6 ml 0.05 M
sulfuric acid each. The combined aqueous phases were
filtered through glass-wool onto a solid phase extraction
cartridge (SCX-SPE, 6 ml/500 mg, Phenomenex, As-
chaffenburg, Germany). Pre-conditioning of the SPE car-
tridge was performed by 6 ml methanol and 6 ml 0.05 M
sulfuric acid. The loaded column was first washed with
12 ml deionized water followed by 12 ml methanol. Elu-
tion (flow: 1 ml/min) of the target compounds was
accomplished with 3 9 6 ml ammoniacal methanol (6 ml
NH4OHcc in 100 ml methanol). The eluate was dried under
Recently, an HPLC–ESI–MS/MS method was developed
and validated to quantitate PA in a plant leaf matrix [21].
This method and sample preparation were modified to
analyze PA in a Boraginaceae oil matrix. An HPLC (Ag-
ilent 1200 Series, Waldbronn, Germany) was coupled to a
3200 QTrap mass spectrometer (Applied Biosystems MDS
Sciex, Darmstadt, Germany). A Gemini-NX C18 column
(150 mm 9 2.0 mm, Phenomenex, Aschaffenburg, Ger-
˚
many) with a particle size of 3 lm and a pore size of 110 A
was used for chromatographic separation. The gradient of
two solvents (A: 25 mM ammonium bicarbonate solution,
B: acetonitrile) was used to separate the target compounds
(flow rate 250 ll/min): 0–3 min (95 % A), 3–28 min
(95–0 % A), 28–31 min (0 % A), 31–32 min (0–95 % A),
32–42 min (re-equilibration 95 % A).
Ionization was in the ESI-positive mode. Quantification
of the analytes and internal standard was performed by
using the MRM mode (multi reaction monitoring) with the
following parameters:
123

J Am Oil Chem Soc
Curtain gas 42 psi, ionization voltage 4,500 V, ion
source temperature 450 °C, gas 1:50.0 psi, gas 2:35.0 psi,
collisionally activated dissociation gas: medium, declus-
tering potential 71.0 V, entrance potential 8.0 V, collision
cell exit potential 5.0 V.
For the internal standard ([9,9-2H2]-retronecine-diphth-
alate): 454.0 amu ? 96.0 amu (CEP 18.0, CE 50.0),
454.0 amu ? 122.0 amu(CEP18.0, CE36.0),454.0amu ?
149.0 amu(CEP24.0, CE55.0)and454.0 amu ?306.0 amu
(CEP 24.0, CE 40.0).
Four transitions for the analytes and three transitions for
the internal standard were selected:
For the first analyte (senecionine): 336.0 amu ?
94.0 amu [cell entrance potential (CEP) 19.55, collision
energy (CE) 50.0], 336.0 amu ? 120.0 amu (CEP 19.55,
CE 35.0), 336.0 amu ? 138.0 amu (CEP 19.55, CE 35.0)
and 336.0 amu ? 308.0 amu (CEP 19.55, CE 35.0).
For the second analyte (SenNOx): 352.0 amu ?
94.0 amu (CEP 20.0, CE 60.0), 352.0 amu ? 120.0 amu
(CEP 20.0, CE 40.0), 352.0 amu ? 136.0 amu (CEP 20.0,
CE 40.0) and 352.0 amu ? 324.0 amu (CEP 20.0, CE
35.0).
Results and Discussion
Recently, the EFSA has published an opinion on pyrrolizidine
alkaloids in food and feed [15]. A BMDL10 for an excess
cancer risk of 70 lg/kg b.w./day was calculated for induction
of liver hemangiosarcomas by lasiocarpine in male rats. A
margin of exposure (MOE) approach by using the BMDL10
for excess cancer risk in male rats was chosen as the reference
point for comparison. The EFSA concluded that a MOE of at
least10,000indicatedthatPAdosesofupto0.007 lg/kgb.w./
day are unlikely to be of concern for cancer risk [15].
For the internal standard (seneciphylline): 334.0 amu ?
94.0 amu (CEP 19.5, CE 50.0), 334.0 amu ? 120.0 amu
(CEP 19.5, CE 35.0) and 334.0 amu ? 138.0 amu (CEP
19.5, CE 35.0).
So far, only the PA content of phytopharmaceuticals is
regulated in several European countries, e.g. the internal
use of such preparations is limited in Germany to levels of
1 lg total 1,2-unsaturated PA per day (if the period
exceeds 6 weeks, then 0.1 lg PA/day) or the Netherlands
1 lg PA/kg or 1 lg PA/l [22]. In the case of the refined E.
plantagineum oil, approved as a novel food ingredient
under EU Regulations [11], this product was specified to
contain \4 lg PA/kg oil, which was at the same time the
limit of detection (LOD) of the TLC method applied.
Our first approach was to test the influence of the pH of
the washing solution in reducing the PA content of seed oil.
Since the available amounts of L. squarrosa seed oil are
limited at the moment, we initiated our preliminary
experiments using sunflower oil. This matrix was spiked
with a mixture of senecionine and SenNOx (ratio 1:1,
134 lg/kg total PA) to cover for the polarity difference of
both naturally occurring PA forms. As shown in Fig. 1a,
there is a clear increase in PA reduction at lower pH values
and the efficiency can be enhanced if the washing is
repeated (Fig. 1b; shown for pH 2.2). Starting with Sen/
SenNOx levels of 134 lg/kg oil, which is close to natural
occurring levels in crude seed oils of Boraginaceous plants,
it is obvious that washing will reduces the PA levels of
such products. Furthermore, lower pH values will promote
alkaloid protonation, and hence increase its polarity and
water solubility. The best result was obtained at pH 2.2.
Using one washing step under these conditions reduced the
PA level to 1.6 %. Increasing the numbers of washings at
low pH-values is an additional option to further reduce the
PA levels. It was found that three washings at pH 2.2 can
bring the PA level down to 0.08 % (Fig. 1b).
Analyst 1.4.2 Software (Applied Biosystems MDS
Sciex, Darmstadt, Germany) was used for data analysis and
integration.
HPLC–ESI–MS/MS Method for Lab and Pilot Scale
Refinement of Lappula squarrosa oil and Commercial
Boraginaceous Plant Seed Oils
The same HPLC–ESI–MS/MS system was used as
described above. Here, a DiscoveryÒ HS F5 column
(150 mm 9 2.1 mm, particle size 3 lm; Sigma Aldrich,
Seelze, Germany) including a pre-column cartridge of the
same material was used. The following gradient of two
solvents (A: 0.1 M formic acid in water, B: acetonitrile)
was applied with a flow rate of 250 ll/min: 0–3 min (95 %
A), 3–13 min (95–0 % A), 13–18 min (0 % A), 18–19 min
(0–95 % A), 19–29 min (re-equilibration: 95 % A).
Ionization was in the ESI-positive mode. Quantification
of the analytes and internal standard was performed by
using the MRM mode with the following parameters:
Curtain gas 20 psi, ionization voltage 5,500 V, ion
source temperature 475 °C, gas 1:50.0 psi, gas 2:70.0 psi,
collisionally activated dissociation gas: high, declustering
potential 71.0 V, entrance potential 8.5 V, collision cell
exit potential 4.0 V.
Four transitions for the analytes and internal standard
were selected:
For the analytes (retronecine/heliotridine-diphthalate):
452.0 amu ? 94.0 amu (CEP 18.0, CE 50.0), 452.0 amu ?
120.0 amu (CEP 18.0, CE 36.0), 452.0 amu ? 149.0 amu
(CEP 18.0, CE 55.0) and 452.0 amu ? 304.0 amu (CEP
24.0, CE 40.0).
These first model experiments were useful to establish
our analytical method. It allowed us to test and optimize
our sample extraction procedure. The resulting extracts
123

J Am Oil Chem Soc
Fig. 1 Reduction of the PA
content of PA-spiked sunflower
oil performing washing steps at
a different pH values or
(a) 160
140
120
100
80
b multiple washings at pH 2.2
60
40
20
0
pH 7.0
pH 6.0
pH 5.0
pH 4.0
pH 3.0
pH 2.2
(b)
20
16
12
8
4
0
1. wash
2. wash
3. wash
4. wash
Fig. 2 Reduction of the
senecionine/senecionine-N-
oxide content of PA spiked
Lappula squarrosa oil at
different pH values (n = 3)
80
70
60
50
40
30
20
10
0
without
treatment
pH 7.0
pH 6.0
pH 5.0
pH 4.0
pH 3.0
pH 2.2
were analyzed with
a
selective HPLC–ESI–MS/MS
more complex and difficult matrix. In addition, L. squar-
rosa oil is characterized by large amounts of PUFA. Hence,
it was of interest to verify the results obtained with sun-
flower oil to find the most efficient and gentle conditions
for L. squarrosa refinement. L. squarrosa oil was spiked
method to quickly track the fate and the behavior of Sen/
SenNOx under such model conditions.
Compared to sunflower oil, crude L. squarrosa oil has a
dark greenish/brown color and intense odor representing a
123

J Am Oil Chem Soc
Fig. 3 Scheme of the sample preparation and HPLC–ESI–MS/MS sum parameter analysis of the total 1,2-unsaturated PA content in oil and oil
seeds
with known amounts of Sen/SenNOx and seneciphylline
was used as an internal standard to quantify Sen/SenNOx
and to calculate recovery rates of the PA in the matrix
crude L. squarrosa oil. All three PA are common for
Senecio spp. (Asteraceae) but do not occur in L. squarrosa
(Boraginaceae). As shown in Fig. 2, washing at pH 6–7 can
bring down the PA level to 13–15 %, but again only
washing at pH 2.2 efficiently reduced the PA content down
to 0.1 % of the starting value (Fig. 2).
So far, the PA of L. squarrosa are not known and
standards for target analysis are not available. Hence, we
decided to apply a PA sum parameter method to quantify
the total PA content in seed oil at different stages of the
refinement. After extraction and concentration on SCX-
SPE cartridges, all PA/PANO of the sample are simulta-
neously reduced to the corresponding necine base diols
(see Fig. 3). Those diols showed unfavorable characteris-
tics in HPLC chromatography and ESI–MS/MS. Hence,
prior to HPLC–ESI–MS/MS these diols had to be con-
verted into the corresponding phthalic acid diesters using a
phthalic anhydride/pyridine derivatization. Using this
approach, we were able to detect all 1,2-unsaturated
While the Sen/SenNOx model experiments were helpful
for establishing sample preparation/PA concentration and
pinpointing crucial steps in oil refinement, they lacked
specificity in terms of the PA relevant for L. squarrosa.
123

J Am Oil Chem Soc
Fig. 4 Enhanced product ion
(EPI)-ESI–MS/MS spectra of
a the derivatized analytes
(retronecine-diphthalate/
(a)
100
120.0
heliotridine-diphthalate) and
b the corresponding derivatized
deuterated internal standard
([9,9-2H2]-retronecine-
HO
O
1
O
OH
4
9
O
N
O
H
O
O
30
diphthalate) including proposed
structures for fragment ions
4
+
H
1
)
3
N
H
2
8
8
(
+
H
6
)
50
+
+
[M H]
=
.
Mr 452 1
149.0
N
C
H
3
94.0
138.1
140
286.0 304.1
250.0
452.0
460
60
100
180
220
260
300 340
380
420
m/z
(b)
100
122.0
H
O
O
1
4
9
H
O
O
O
O
D
O
O
D
30
6
+
H
)
1
2
4
N
0
88
(
+
H
)
H
50
+
+
M H
]
[
=
.
Mr 454 1
149.0
96.0
100
140.0
454.0
460
306.1
288.1
60
140
180
220
260
m/z
300 340
380
420
retronecine/heliotridine-type PA in the form of a single
sum parameter expressed as retronecine equivalents (RE).
Accurate quantification was achieved using 7-O,9-O-di-
butyroyl-[9,9-2H2]-retronecine as a stable isotope-labeled
internal standard (Figs. 3, 4) [21].
MS method allowed the unequivocally determination of PA
levels at all different stages of the refinement process,
achieving a LOD and a limit of quantification (LOQ) of 0.02
and 0.05 lg RE/kg, respectively. Besides the mass spectro-
metric dependability (one quantifier transition, three quali-
fier transitions), the LOQ is 80–400 times lower than the
LOD of the TLC method (LOD 4 lg/kg) applied in the novel
food application of E. plantagineum oil or the LOD reported
for a GC–MS approach (20 lg/kg) used to quantify the PA
content in crude and processed Borage oil, respectively [24,
25]. This increase in sensitivity is desirable in the course of
the toxicological importance of the PA and the EFSA
assessment that only PA doses up to 0.007 lg/kg b.w./day
are unlikely to be of concern for cancer risk [15].
The analytes (retronecine-diphthalate/heliotridine-diph-
thalate) co-elute together with the internal standard deriva-
tive [9,9-2H2]-retronecine-diphthalate, covering for possible
matrix effects. This is especially true when larger quantities
of oil are used in sample preparation to achieve low detection
limits. Quantification was achieved by monitoring analyte
(retronecine/heliotridine-diphthalate) specific MRM transi-
tions and specific transitions for the internal standard,
respectively. The application of this new HPLC–ESI–MS/
123

J Am Oil Chem Soc
Applying this sensitive sum parameter method we were
able to explore the impact of additional refinement steps
and the concurrent PA reductions. Figure 5 summarizes the
two extremes of the experimental setup, (1) neutral con-
ditions pH 7, being the most gentle washing step and (2)
pH 2.2 representing the harshest conditions but highest
efficiency in terms of PA reduction, both combined with an
additional neutralization and deodorization step under
laboratory conditions. At neutral pH, this combination
showed a PA reduction from 33 lg RE/kg (crude L.
squarrosa oil) down to 0.4 lg RE/kg, while the same
procedure at pH 2.2 resulted in a ten times lower PA
content right at the LOQ (0.04 lg RE/kg).
L. squarrosa oil is still a rare product whose market
potential is still being explored. At this point it was decided
that one crucial milestone for future development is food
safety. Hence, a pilot scale refinement process (50 kg
batch) was initiated to investigate a combination of steps
which would be reasonable to produce high quality food
grade L. squarrosa oil with the lowest possible PA content.
The overall process was performed in a mini plant allowing
an easy scale-up of the process to produce economic rel-
evant amounts of L. squarrosa oil. The starting material
was L. squarrosa seeds showing 16 mg RE/kg. The
resulting crude L. squarrosa oil contained 29 lg RE/kg
which was consecutively reduced to 0.9 lg RE/kg
Fig. 5 Comparison of the
reduction of total 1,2-
unsaturated PA content of crude
Lappula squarrosa oil using
washing and deodorization at
pH 7 and 2.2 (n = 3)
Fig. 6 Monitoring of the total
1,2-unsaturated PA content at
different stages of a pilot scale
refinement of crude Lappula
squarrosa oil and the starting
seed material (n = 3)
123

J Am Oil Chem Soc
valuable tools for an efficient reduction in PA found in
crude Boraginaceous seed oils. On the way to exploring the
possible market potential of L. squarrosa oil, a pilot scale
refinement was established with the aim of the best pos-
sible PA reduction for food safety reasons but without
sacrificing the future options for scale up for a commercial
process. The study was finalized by the analysis of com-
mercially available Boraginaceous seed oil products and
food supplements using the newly developed HPLC-ESI–
MS/MS method to survey the PA content of those niche
products.
Table 1 Total PA content of
commercially available
Boraginaceous plant oil/oil
products in retronecine
equivalents (RE)
Sample
lg RE/kg oil
BO_1
BO_2
BO_3
BO_4
BO_5
EO_1
EO_2
EO_3
EOP_1
EOP_2
0.08
ND
0.5
ND
ND
ND
ND
ND
ND
0.6
BO, Borago officinalis oil; EO, Echium
spp. oil; EOP, Echium spp. oil product;
ND, not detected
Acknowledgments The authors express their gratitude to Astrid
Kupfer (Exsemine) for her helpful assistance.
(washing at pH 2.2) down to 0.12 lg RE/kg (after
bleaching). At the next stage (deodorization) PA were
below the LOQ (0.05 lg RE/kg) (Fig. 6). Using this pro-
cedure, we were able to demonstrate that PA levels of L.
squarrosa oil can be significantly reduced to levels below
any toxicological concern.
Conflict of interest The authors have no conflict of interest to
report.
References
Finally, we used our established method to determine
the PA content of commercially available Boraginaceous
seed oil products. Overall ten different products, including
pure seed oils from Echium spp. or B. officinalis and cap-
suled food supplements containing inter alia Echium spp.
oil were analyzed. The analytical results are summarized in
Table 1. Three out of these ten products were tested PA
positive. The PA content was ranging from not detected to
0.6 lg RE/kg product (corresponding to 1.2 lg 1,2-unsat-
urated PA/kg using a conversion factor of 2). Ironically,
one of the positive samples was labeled as ‘‘produced by
using PA free Borago seeds’’ (sample BO_3; Table 1).
Assuming a daily dose of 10 g of such products would
be reflected in 6 ng RE/day (assuming a conversion factor
of two to calculate for the total PA content) this would be
*12 ng PA/day which most likely gives no reason for
toxicological concerns. However, in terms of preventive
consumer protection and other possible sources of PA
intake into the food chain, our studies on the refinement of
L. squarrosa oil and the PA negative commercial products
demonstrate that the PA content can be reduced to non-
detectable (LOD 0.02 lg RE/kg) using appropriate refine-
ment conditions for such oil seeds/products.
1. EFSA Panel on Dietetic Products, Nutrition, and Allergies (NDA)
(2010) Scientific opinion on dietary reference values for fats,
including saturated fatty acids, polyunsaturated fatty acids,
monounsaturated fatty acids, trans fatty acids, and cholesterol.
EFSA J 8:1461–1568. http://www.efsa.europa.eu/de/efsajournal/
pub/1461.htm. Accessed 12 Sept 2013
2. Harizi H, Corcuff JB, Gualde N (2008) Arachidonic-acid-derived
eicosanoids: roles in biology and immunopathology. Trends Mol
Med 14:461–469
3. Komprda T (2012) Eicosapentaenoic and docosahexaenoic acids
as inflammation-modulating and lipid homeostasis influencing
nutraceuticals: a review. J Funct Food 4:25–38
¨
4. DEG, Evidenzbasierte Leitlinie, Fettkonsum und Pravention
¨
¨
ausgewahlter ernahrungsmitbedingter Krankheiten, 2006. http://
www.dge.de/pdf/ws/ll-fett/DGE-Leitlinie-Fett-11-2006.pdf.
Accessed 12 Sept 2013
5. SOFA Database on Seed Oil Fatty Acids. http://sofa.mri.bund.de/
. Accessed 12 Sept 2013
6. James MJ, Ursin VM, Cleland LG (2003) Metabolism of steari-
donic acid in human subjects: comparison with the metabolism of
other n-3 fatty acids. Am J Clin Nutr 77:1140–1145
7. Whelan J (2009) Dietary stearidonic acid is a long chain (n-3)
polyunsaturated fatty acid with potential health benefits. J Nutr
139:5–10
8. Whelan J, Gouffon J, Zhao Y (2001) Effects of dietary steari-
donic acid on biomarkers of lipid metabolism. J Nutr. doi:10.
3945/jn.111.149138, 1S–5S
9. Surette ME (2013) Dietary omega-3 PUFA and health: steari-
donic acid-containing seed oils as effective and sustainable
alternatives to traditional marine oils. Mol Nutr Food Res
57:748–759
10. Hammond BG, Lemen JK, Ahmed G, Miller KD, Kirkpatrick J,
Fleeman T (2008) Safety assessment of SDA soybean oil: results
of a 28-day gavage study and a 90-day/one generation repro-
duction feeding study in rats. Regul Toxicol Pharmacol
52:311–323
11. Commission Decision (2008) Authorising the placing on the
market of refined echium oil as novel food ingredient under
Regulation (EC) No. 258/97 of the European Parliament and of
the Council (notified under document number C(2008) 3049;
(2008/558/EC). http://eur-lex.europa.eu/LexUriServ/LexUriServ.
Conclusions
The presented results pinpoint crucial steps in the refine-
ment of L. squarrosa seed oil for the most effective
reduction of the accompanying PA. We adopted an HPLC–
ESI–MS/MS sum parameter method which enabled us to
measure trace levels of PA in oil matrices. It was demon-
strated that washing at low pH and deodorization are
123

J Am Oil Chem Soc
do?uri=OJ:L:2008:180:0017:0019:EN:PDF. Accessed 12 Sept
2013
¨
12. Ozcan T (2008) Analysis of the total oil and fatty acid compo-
19. Guil-Guerrero JL (2007) Stearidonic acid (18:4n-3): metabolism,
nutritional importance, medical uses and natural sources. Eur J
Lipid Sci Technol 109:1226–1236
sition of seeds of some Boraginaceae taxa from Turkey. Plant
Syst Evol 274:143–153
20. Khatmulina LI, Fedorov NI, Ermolaeva NA, Galkin EG, Yunu-
sov MS (2012) Polyunsaturated fatty acids from several plant
species of the family Boraginaceae. Chem Nat Compd
48:361–366
21. Cramer L, Schiebel H-M, Ernst L, Beuerle T (2013) Pyrrolizidine
alkaloids in the food chain: development, validation, and appli-
cation of a new HPLC–ESI–MS/MS sum parameter method.
J Agric Food Chem 61:11382–11391
22. Kempf M, Reinhard A, Beuerle T (2010) Pyrrolizidine alkaloids
(PAs) in honey and pollen-legal regulation of PA levels in food
and animal feed required. Mol Nutr Food Res 54:158–168
23. Cramer L, Beuerle T (2012) Detection and quantification of
pyrrolizidine alkaloids in antibacterial medical honeys. Planta
Med 78:1976–1982
24. Mierendorff HJ (1995) Determination of pyrrolizidine alkaloids
by thin-layer-chromatography in the oil of seeds of Borago offi-
cinalis L. Lipid 97:33–37
´
´
13. Guil-Guerrero JL, Rincon-Cervera MA, Gomez-Mercado F, Ra-
mos-Bueno RP, Venegas-Venegas E (2013) New seed oils of
Boraginaceae rich in stearidonic and gamma-linolenic acids from
the Maghreb region. J Food Compos Anal 31:20–23
14. Hartmann T, Witte L (1995) Pyrrolizidine alkaloids: chemical,
biological and chemoecological aspects. In: Pelletier SW (ed)
Alkaloids: chemical and biological perspectives, vol 9. Pergamon
Press, Oxford, pp 155–233
15. EFSA (2011) Scientific opinion on pyrrolizidine alkaloids in food
and feed. EFSA J 9. doi:10.2903/j.efsa. http://www.efsa.europa.
eu/de/efsajournal/pub/2406.htm. Accessed 12 Sept 2013
16. Hegi G (1927) Illustrierte Flora von Mittel-Europa. Band V/3,
Carl Hanser Mu¨nchen, 2138–2142 and 2220–2228
17. Craig BM, Bhatty MK (1964) A naturally occurring all-cis
6,9,12,15-octadecatetraenoic acid in plant oils. JAOCS
41:209–211
¨
25. Wretensjo I, Karlberg B (2003) Pyrrolizidine alkaloid content in
18. Gunstone FD, Hamilton RJ, Padley FB, Qureshi MI (1965)
Glyceride studies. V. The distribution of unsaturated acyl groups
in vegetable triglycerides. JAOCS 42:965–970
crude and processed borage oil from different processing stages.
JAOCS 80:963–970
123