Analysis of reaction products of food contaminants and ingredients: Bisphenol A diglycidyl ether (BADGE) in canned foods

Article abstract of DOI:10.1021/jf904160a

Bisphenol A diglycidyl ether (BADGE) is an epoxide that is used as a starting substance in the manufacture of can coatings for food-contact applications. Following migration from the can coating into food, BADGE levels decay and new reaction products are

Full text of DOI:10.1021/jf904160a

J. Agric. Food Chem. 2010, 58, 4873–4882 4873
DOI:10.1021/jf904160a
Analysis of Reaction Products of Food Contaminants and
Ingredients: Bisphenol A Diglycidyl Ether (BADGE) in Canned
Foods
†
L
EON
C ,
OULIER,*,† EMMA L. BRADLEY,§ RICHARD C. BAS,† KITTY C. M. VERHOECKX
§
§
MALCOLM
D
RIFFIELD,§ NICK
HARMER
,
AND
L
AURENCE
C
ASTLE
†TNO Quality of Life, Utrechtseweg 48, 3704 HE Zeist, The Netherlands, and § The Food and
Environment Research Agency, York Y041 1LZ, United Kingdom
Bisphenol A diglycidyl ether (BADGE) is an epoxide that is used as a starting substance in the
manufacture of can coatings for food-contact applications. Following migration from the can coating
into food, BADGE levels decay and new reaction products are formed by reaction with food
ingredients. The significant decay of BADGE was demonstrated by liquid chromatographic (LC)
analysis of foodstuffs, that is, tuna, apple puree, and beer, spiked with BADGE before processing
and storage. Life-science inspired analytical approaches were successfully applied to study the
reactions of BADGE with food ingredients, for example, amino acids and sugars. An improved mass
balance of BADGE was achieved by selective detection of reaction products of BADGE with low
molecular weight food components, using a successful combination of stable isotopes of BADGE
and analysis by LC coupled to fluorescence detection (FLD) and high-resolution mass spectrometric
(MS) detection. Furthermore, proteomics approaches showed that BADGE also reacts with peptides
(from protein digests in model systems) and with proteins in foods. The predominant reaction center
for amino acids, peptides, and proteins was cysteine.
KEYWORDS: Bisphenol A diglycidyl ether (BADGE); canned foods; food packaging; reaction pro-
ducts; food ingredients; stable isotopes
INTRODUCTION
1 mg/kg for the sum of the migration of BADGE HCl, BAD-
3
GE 2HCl, and BADGE HCl H2O (2). The migration of
3
3
3
Most food and beverage cans are internally coated to protect
the foodstuff from the metal and, vice versa, to prevent metal
corrosion occurring with aggressive food ingredients. The domi-
nant type of coating for food and beverage cans is epoxy-phenolic
based on bisphenol A and epichlorohydrin. These two starting
substances are reacted together to form a low molecular weight
prepolymerized resin, including bisphenol A diglycidyl ether
(BADGE, Figure 1) and higher oligomers. The resin is then used
to coat the metal, which is cured to form a hard polymeric
protective coating.
The migration of BADGE from food contact materials has
received a lot of attention in the past. The main focus then was on
applications where BADGE was used not as a starting substance,
for example, for epoxy-phenolic coatings, but was used as an
additive (as a stabilizer) in certain PVC organosol can coatings. In
response to these findings, a specific EU Directive on BADGE
was prepared limiting the migration of BADGE and its deriva-
tives (BADGE, BADGE H2O, BADGE 2H2O, BADGE HCl,
BADGE and its derivatives is normally determined using so-
called food simulants. These food simulants were introduced to
mimic certain types of food; for example, olive oil represents fatty
food (3). More than 50 papers have been published on this topic,
mostly concerned with methods of analysis and surveillance, (see,
e.g., ref4-8). Most often, reversed-phase liquid chromatography
using C8 or C18 columns in combination with a mixture of
acetonitrile/water as mobile phase and coupled to fluorescence
detection (FLD) is used for the analysis of BADGE and related
products. Basically, if the migration of BADGE and its known
derivatives into food simulants is below the migration limit, the
food contact material containing BADGE is in accordance with
the current legislation. Although compliance is usually demon-
strated by the analysis of food simulants exposed to the can
coating, the concentrations of BADGE and its derivatives can
also be measured in foodstuffs. From a practical point of view it is
not obligatory to measure these substances in real food products.
However, some studies on BADGE in food products have been
carried out. The concentrations detected in the foodstuffs gave
low recoveries for BADGE that could not be explained by the
formation of its known derivatives (9, 10). It was suggested that
migrated BADGE reacted with food components. A few studies
found evidence of the reactivity of BADGE toward food com-
ponents such as amino acids and proteins (10-12). Recently,
3
3
3
BADGE 2HCl, and BADGE HCl H2O) to 1 mg/kg food or
3
3
3
food simulant (1). This was later updated in light of new toxi-
cological data, to a restriction of 9 mg/kg for the sum of the
migration of BADGE, BADGE H2O, and BADGE 2H2O and
3
3
*Author to whom correspondence should be addressed. [telephone
(31) 30 694 48 88; fax (31) 30 694 4894; e-mail leon.coulier@tno.nl].
© 2010 American Chemical Society
Published on Web 03/24/2010
pubs.acs.org/JAFC

4874 J. Agric. Food Chem., Vol. 58, No. 8, 2010
Coulier et al.
Figure 1. Structures of BADGE, BADGE H2O, BADGE 2H2O, and BADGE-d14.
3
3
Table 1. Recovery of BADGE and Its Hydrolysis Products BADGE H2O
and BADGE 2H2O in Various Foods and Drinks after Spiking (20 mg/kg),
Canning, and Storing
Petersen et al. (10) investigated the reaction of BADGE with
proteins in more detail. The analysis of BADGE and possible
reaction products in extracts of complex food matrices is not
straightforward, leading to complex chromatograms in which it is
difficult to identify the BADGE-related products (13).
To deal with the challenges of analyzing BADGE-related
reaction products, which can differ widely in molecular weight,
polarity, and chemical structure, in complex food matrices,
sophisticated analytical approaches are necessary. In this study
analytical technology inspired by life-science applications, such as
the use of isotopically labeled compounds and peptide/protein
analysis, is used to study the stability of BADGE and the
formation of reaction products with food ingredients.
3
3
sample
storage
recovery (%)
tuna in oil
3 weeks at room temperature
12 weeks at room temperature
3 weeks at room temperature
þ 12 weeks at 50 °C
25
0
2
apple puree
3 weeks at room temperature
12 weeks at room temperature
3 weeks at room temperature
þ 12 weeks at 50 °C
15
12
17
ale
3 weeks at 40 °C
3 weeks at 40 °C
3 weeks at 40 °C
26
5
lager
stout
MATERIALS AND METHODS
8
Samples. Five food and beverage samples were selected for investiga-
tion. Two canned food samples, tuna in sunflower oil and apple puree,
were prepared by a commercial food packer, and three beverage samples,
an ale, a stout, and a lager, were purchased from retail outlets in northern
England. For each foodstuff 10 control samples were prepared to which no
BADGE was added and 50 samples were prepared and spiked with
BADGE at a nominal concentration of 20 mg/kg. The same procedure
was repeated at a later stage in the study in which both BADGE and
deuterated BADGE were spiked into the food samples.
Canned Foods. Tuna in sunflower oil and apple puree were packed in
metal cans lacquered internally with a standard epoxy-phenolic coating. For
epoxy-phenolic coatings, BADGE is used as a starting substance, and it is
intended to be used-up in the polymerization reaction. These samples were
recanned, that is, they were obtained from cans, then “spiked” with BADGE,
and then recanned. Control samples, canned water, and sunflower oil were also
prepared. The canned samples were heated in a static steam sterilizer. For tuna,
the heating time was 50 min at 122 °C steam temperature, and for apple puree,
the low pH (pH 3.7) required a lower processing temperature, that is, a heating
time of 38 min at 105 °C steam temperature. The samples were then stored at
room temperature, and after 3 weeks, half of the cans were transferred to an
oven set at 50 °C and the remainder were left at room temperature.
of water-free acetonitrile. The resulting solution was collected and dried at
60 °C and 150 mbar using a rotary evaporator. In this way BADGE-d14
(Figure 1) was synthesized with a yield of 85%.
Analysis of (Spiked) Food Samples. The content of each can/bottle
was homogenized, and duplicate portions (5 g) were taken. A further two
5 g portions were taken from each nonspiked food sample to determine
analytical recovery and were fortified with BADGE, BADGE H2O, and
3
BADGE 2H2O (Sigma-Aldrich) each at a concentration of 20 mg/kg. The
3
samples were extracted into acetonitrile (5 mL) by shaking for 18 h at room
temperature. Each sample was centrifuged and the supernatant passed
through a 0.45 μm filter. Calibration solutions (0-25 mg/L) were prepared
and run alongside the samples.
Sample extracts were analyzed by liquid chromatography with fluore-
scence detection (LC-FLD) and/or mass spectrometric detection (LC-
MS). LC-FLD analyses were performed on a Hewlett-Packard 1100 series
LC equipped with an automatic degasser, binary pump, autosampler, and
fluorescence detector (Agilent Technologies, Palo Alto, CA). Chromato-
graphic separation was carried out using a Sunfire C18 column with
dimensions of 150 mm ꢀ 3.0 mm i.d., 3.5 μm (Waters, Milford, MA), with
a mobile phase gradient from 80% water þ 0.1% trifluoroacetic acid (A)
to 100% acetonitrile þ 0.1% trifluoroacetic acid (B) with a flow of 0.4 mL/
min using a mobile phase gradient from 80 to 50% A in 22 min followed by
8 min at 50% A and rinsing of the column with 100% B for 5 min. Injection
volume was 10 μL. The fluorescence detector was set at excitation and
emission wavelengths of 275 and 305 nm, respectively. Quantification was
Bottled Beers. Three sets of beer samples were prepared. Bottled beers
were opened in a carbon dioxide-filled glovebag (to maintain carbona-
tion). The beers were spiked with BADGE and then recapped using fresh
crown caps. The samples were stored at 40 °C.
Preparation of Deuterated BADGE. The protocol for the synthesis
of labeled BADGE from bisphenol A-d16 was adapted from that of bis-
phenol F diglycidyl ether (BFDGE) (14). In short, 1 g of bisphenol A-d16
(98 atom % D, Sigma-Aldrich Chemie, Zwijndrecht, The Netherlands)
was mixed with 0.65 g of KOH and 50 mL of ethanol. The mixture
was evaporated to dryness using a rotary evaporator (60 °C, 150 mbar)
followed by drying under vacuum at 60 °C overnight. Next, 30 mL of
water-free acetonitrile (50 g of Na2SO4 þ 200 mL of acetonitrile) and 9 mL
of epibromohydrin (Sigma-Aldrich) were added to the dried potassium
phenolate. The mixture was stirred for 30 min at room temperature and for
1 h at 50 °C and then refluxed for 10 min. Next, the mixture was cooled to
room temperature and filtered. The precipitate was washedwith 2 ꢀ 30 mL
carried out using solvent standards of BADGE, BADGE H2O, and
3
BADGE 2H2O (0-25 mg/L).
3
LC-FT-MS analyses were performed on an LTQ Fourier Transform
(FT) linear ion trap system consisting of a Surveyor AS autosampler and a
Surveyor MS pump, equipped with an LTQ LT-1000 FT mass detector
with an Opton ESI probe (Thermo Fisher Scientific, San Jose, CA).
Chromatographic separation for LC-MS was carried out using a Sunfire
C18 column with dimensions of 150 mm ꢀ 3.0 mm, 3.5 μm (Waters), with a
mobile phase gradient from 80% 5 mM ammonium acetate adjusted to pH 5
with acetic acid (A) to 50% 2.5 mM ammonium acetate in acetonitrile (B) in
22 min followed by 8 min at 50% A/50% B and rinsing of the column with

Article
J. Agric. Food Chem., Vol. 58, No. 8, 2010 4875
Figure 2. (A) LC-FLD chromatograms of BADGE (- - -) and a reaction mixture of BADGE and glucose after reflux for 2 h at 100 °C (;); (B) LC-FT-MS
chromatogram of reaction product of BADGE and glucose with corresponding mass spectrum.
100% B for 5 min with a flow of 0.4 mL/min. Injection volume was 10 μL.
Mass detection was carried out using electrospray ionization (ESI) in
the positive mode (m/z 100-1200). Mass accuracy was generally better
than 2 ppm. Calibration and optimization were carried out using solvent
standards of BADGE, BADGE H2O, and BADGE 2H2O.
BADGE-Monosaccharide Adducts. The reaction between
BADGE and glucose was carried out by refluxing an aqueous solution
containing 50 μg of BADGE, 65 μg of BADGE-d14, and 500 mg of
glucose in 5 mL of water for 2 h at 100 °C. The reaction mixture was
analyzed by LC-FLD and LC-MS.
BADGE-Amino Acid Adducts. BADGE (20 μg) was reacted with
the 250 μg of the amino acid methionine, cysteine, histidine, lysine, or
tyrosine in 25 mL of water for 30 min at 100 °C and also at room
temperature. The reaction mixtures were analyzed by LC-FLD and
LC-MS.
BADGE-Protein Adducts. Reaction of BADGE with insulin
(Sigma-Aldrich) and bovine serum albumin (BSA, Sigma-Aldrich) was
carried out for 30 min at 100 °C before and after digestion. To this purpose
100 μL of BADGE solution (∼1 mg/mL) in acetonitrile was evaporated
to dryness and redissolved in 100 μL of protein solutions (1 mg/mL) or
3
3

4876 J. Agric. Food Chem., Vol. 58, No. 8, 2010
Coulier et al.
Table 2. Examples of Reaction Products of BADGE with Food Ingredients Identified by High-Resolution LC-FT-MS
elemental composition after
subtraction of adducts and BADGEa
reaction
observed m/z
elemental composition
reaction product
BADGE þ glucose
BADGE þ cysteine
BADGE þ cysteine
BADGE þ methionine
BADGE þ methionine
BADGE þ methionine
BADGE þ lysine
[M þ NH4]þ = 556.2748
[M þ H]þ = 480.2049
[M þ H]þ = 583.2141
[M þ NH4]þ = 424.2145
[M þ NH4]þ = 454.2117
[M þ H]þ = 508.2359
[M þ H]þ = 505.2905
C27H42O11N1
C24H34O7N1S1
C27H39O8N2S2
C22H34O5N1S1
C23H36O4N1S2
C26H38O7N1S1
C27H41O7N2
C6H12O6
BADGE H2O glucose
3
3
C3H7NO2S1
C6H14N2O4S2
SCH3
BADGE H2O Cys
3
3
BADGE 2Cys
3
BADGE H2O SCH3
3
3
BADGE 2SCH3
S2C2H6
C5H11O2N1S1
C6H14O2N2
3
3
3
BADGE H2O.Met
BADGE H2O Lys
3
a BADGE = C21H24O4, BADGE H2O = C21H26O5, BADGE 2H2O = C21H28O6.
3
3
100 μL of digest solution. Insulin, a low molecular weight protein, was
reacted with BADGE for 30 min at 100 °C and subsequently directly
infused into the MS source.
Trypsin Digestion. To this purpose 100 μL of aqueous protein
solutions (∼10 mg/mL), 800 μL of 80 mM ammonium bicarbonate,
50 μL of 5 mM calcium chloride, and 20 μL of 2 mM dithiothreitol were
mixed, and 20 μL of a trypsin solution (∼1 mg/mL, Sigma-Aldrich) was
added, resulting in an enzyme/protein ratio of 1:100. Digestion was carried
out overnight at 37 °C.
Pronase Digestion. To this purpose 500 μL of aqueous protein
solutions (∼1 mg/mL), 500 μL of 100 mM ammonium bicarbonate, and
1.6 μL of 0.34 mM calcium chloride were mixed, and 100 μL of Pronase
solution (∼1 mg/mL, Sigma-Aldrich) was added, leading to an enzyme/
protein ratio of 1:100. Digestion was carried out overnight at 37 °C.
Analysis of Proteins/Peptides. Tryptic peptides were analyzed by
LC-MS/MS on an LTQ linear ion-trap system (Thermo Fisher Scientific)
using an Inertsil ODS 2 column with dimensions of 100 mm ꢀ 3 mm
(GL Sciences, Torrance, CA) and a mobile phase gradient from 95%
water þ 0.1% formic acid to 100% water þ acetonitrile (20:80) þ 0.1%
formic acid in 20 min with a flow of 0.4 mL/min. Mass detection was
carried out using ESI in the positive mode (m/z 150-2000). Mass spectra
were collected in the full scan mode and in data-dependent mode.
The reaction of BADGE with glucose resulted in the formation
of an additional feature observed with LC-FLD (Figure 2A), and
using high mass resolution LC-FT-MS, this feature could be well
explained by a reaction product of BADGE and glucose, that is,
BADGE H2O glucose (Figure 2B and Table 2). Hence, it could
be concluded that BADGE reacts with glucose and therefore it is
expected that it will react with other monosaccharides in the
same way. This is in contrast to the findings of Petersen et al. (10),
who did not find reaction products of BADGE and glucose.
These authors had similar reaction conditions and analytical
methods, and hence the discrepancy with respect to the presence
or absence of BADGE-glucose reaction products cannot be
easily explained.
3
3
Reaction in model solutions was also carried out with amino
acids. Reaction of BADGE with cysteine resulted in two main
additional peaks observed with LC-FLD that could be identi-
fied by LC-FT-MS as BADGE H2O Cys and BADGE 2Cys
3
3
3
(Figure 3 and Table 2). Similar reaction products were found for
lysine and methionine (Table 2). Although different amino acids
showed reaction products with BADGE, the number and in-
tensity of the peaks differed widely between the different amino
acids. The reaction of BADGE with amino acids increases,
roughly, in the order Tyr < His < Met < Lys < Cys. For
reaction with methionine some additional reaction products were
observed that could be assigned to methylthio-BADGE reaction
products, that is, BADGE H2O SCH3 and BADGE 2SCH3
RESULTS AND DISCUSSION
Levels of BADGE and Its Hydrolysis Products Detected in
Foods. To confirm the disappearance of BADGE in canned foods
and to determine the extent of its disappearance, the levels of
BADGE, BADGE H2O, and BADGE 2H2O (Figure 1) were
3
3
3
3
3
(Table 2). It has been shown that these methylthio derivatives
determined in the foods stored under different conditions using
LC-FLD (Table 1). All packaging materials contained epoxy-
phenolic coatings and not PVC organosol; therefore, chloro-
hydrins could not be formed and were thus not analyzed. The
precision of the method was generally better than 10%.
can be formed from BADGE H2O Met by cleavage and sub-
3
3
sequent transfer of the methylthio group from methionine to
BADGE (10). For all reactions all of the fluorescent response
associated with the BADGE starting material was accounted for
(assuming a 1:1 response). This indicates that the response factors
of BADGE-containing products with LC-FLD are similar,
indicating that the fluorescence response is largely determined
by the BADGE backbone. It also indicates that, as expected, it is
the epoxy groups of BADGE that are the reaction centers
involved and that the fluorescent bisphenol A-like backbone of
BADGE is not disrupted by the reactions.
It can be clearly seen that for all food types and storage conditions
low recoveries were found, varying from 0 to 26%. For tuna in
sunflower oil the recovery was 25% after 3 weeks at room
temperature, whereas after a longer storage time, that is, 12 weeks
at room temperature or 50 °C, nearly all of the BADGE and its
hydrolysis products have disappeared. The recovery in apple puree
was about 15% and remained relatively constant over time. In the
three beer samples the recovery of BADGE, BADGE H2O, and
Profiling of BADGE-Related Products in Foods Using Stable
Isotopes. Following the confirmation that BADGE disappears in
food, which could not be explained quantitatively by the forma-
tion of known hydrolysis products, and the observation that
BADGE is able to react with food ingredients, an analytical
approach was developed to find the reaction products of BADGE
in the different foods. Straightforward analysis using either LC-
FLD or LC-MS is not applicable as these detection methods were
not sufficiently selective to distinguish BADGE reaction products
from other compounds present in food. It was expected that many
different types of compounds with nucleophilic moieties can react
with BADGE, leading to the formation of reaction products with
a wide range of chemical structures and polarities. It can be
expected that the two epoxide groups in BADGE (Figure 1) are
3
BADGE 2H2O ranged from 5 to 26% after 3 weeks at 40 °C. The
3
recoveries in water and sunflower oil were significantly higher, that
is, >80%. These results prove that BADGE “disappears” in the
different foods and that the loss of BADGE cannot be explained by
the formation of the known derivatives, that is, hydrolysis products.
The majority of BADGE disappears in the first 3 weeks of storage.
Reactivity of BADGE. Before the food samples were analyzed
to determine where BADGE had gone, the reactivity of BADGE
toward different food ingredients was investigated. To that
purpose reactions were carried out between BADGE and mono-
saccharides and amino acids, which are compounds present in
many types of foods. The resulting reaction solutions were
analyzed by LC-FLD and high-resolution LC-FT-MS.

Article
J. Agric. Food Chem., Vol. 58, No. 8, 2010 4877
Figure 3. (A) LC-FLD chromatograms of BADGE (- - -) and a reaction mixture of BADGE and cysteine after storage at 100 °C (;); (B) LC-FT-MS
chromatograms and corresponding mass spectra of two main reaction products of BADGE and cysteine.

4878 J. Agric. Food Chem., Vol. 58, No. 8, 2010
Coulier et al.
Figure 4. (A) LC-FLD and (B) LC-FT-MS chromatogram of a standard solution of BADGE and BADGE-d14 and their hydrolysis products.
the main centers of reactivity, resulting in the “addition” of new
molecules to BADGE via the epoxide groups as was confirmed by
the model studies with glucose and amino acids. Isotopically
labeled BADGE, that is, BADGE-d14, was used together with
nonlabeled BADGE. Figure 4 shows LC-FLD and LC-MS
chromatograms of BADGE, BADGE-d14, and their hydrolysis
products in which clearly three doublets of peaks could be
observed. The mass difference of m/z 14 due to the 14 deuterium
atoms resulted in a partial chromatographic separation of the
labeled and nonlabeled BADGE. With LC-MS it can be seen that
the mass difference between the two peaks in the doublets is Δm/z
14, thereby confirming the expected differences between BADGE

Article
J. Agric. Food Chem., Vol. 58, No. 8, 2010 4879
Figure 5. Overlay of LC-FLD chromatograms of direct analysis of stout spiked with BADGE (;) and with BADGE and BADGE-d14 (- - -).
and BADGE-d14. Further experiments showed that BADGE
and BADGE-d14 had sensitivities similar to one another with
both fluorescence and MS detection.
ingredients. Figures 5 and 6 show examples of LC-FLD and
LC-MS chromatograms obtained by direct analysis of stout.
The additional peaks in the sample spiked with BADGE and
BADGE-d14 can be clearly seen in both figures. Moreover, the
sample spiked with BADGE and BADGE-d14 shows the ex-
pected doublets with LC-FLD and the mass difference of m/z 14
with LC-MS. Using this approach, peaks in the LC-FLD
chromatograms were identified as being reaction products of
BADGE. As the exact identities of the additional peaks were not
known, the levels present were quantified by comparing their
peak areas with those derived from the analysis of standard
solutions of BADGE, BADGE H2O, or BADGE 2H2O. It was
shown earlier that BADGE-related products have very similar
fluorescent responses, more or less independent of the food
ingredient. In this way the concentrations of peaks related to
BADGE were added to calculate the recovery of BADGE (and
related products) after storage (Table 3). The resulting recoveries
Mixtures of BADGE and BADGE-d14 were spiked into foods
and were subsequently stored. Analysis of extracts of the food will
result in numerous peaks that can be due to food ingredients
themselves, contaminants introduced due to sample workup, or
BADGE reaction products. However, BADGE and BADGE-
d14 will react with the same food ingredients, and thus reaction
products of BADGE were selectively profiled using the formation
of doublets with LC-FLD and/or the mass difference of Δm/z 14
with LC-MS (Figures 5 and 6).
Analysis of Foods by LC-FLD and LC-FT-MS: Mass Balance
Studies and Identification of Reaction Products. The foods to
which BADGE and BADGE-d14 were spiked prior to storage
were analyzed by both LC-FLD and LC-FT-MS to screen for
reaction products of BADGE with low molecular weight food
3
3

4880 J. Agric. Food Chem., Vol. 58, No. 8, 2010
Coulier et al.
Figure 6. Base peak LC-FT-MS chromatograms of direct analysis of stout spiked with BADGE and BADGE-d14.
of BADGE were significantly higher, especially for lager, stout,
and ale, compared to the recoveries observed earlier in Table 1.
Therefore, it can be concluded that the approach described here,
using isotopically labeled BADGE to identify BADGE-related
products in food (extracts) and fluorescence detection to quanti-
fy, is successful in improving the mass balance of compounds in
complex samples.
LC-FT-MS was also used to identify the reaction products
observed for BADGE in the different foods. From the exact mass
possible elemental compositions were obtained. The number of
possible elemental compositions was reduced by making some
assumptions and using isotope information. The calculated ele-
mental compositions minus that of BADGE (or its hydrolysis
products) was the input for a database search with the aim of
identifying the BADGE reaction products. Some peaks could be
tentatively identified with this approach, for example, reaction
products of BADGE with ethanol and aminophenol in the beer
samples. In apple puree, reaction products of BADGE with
carbohydrates, for example, glucose and/or fructose, could be
identified. This is contrast with Petersen et al. (10), who did not

Article
J. Agric. Food Chem., Vol. 58, No. 8, 2010 4881
Table 3. Recovery of BADGE and Its Reaction Products in the Different
Foods and Drinks after Spiking (20 mg/kg), Canning, and Storing
Table 5. BADGE Adducts of Tryptic Peptides of BSA Found by LC-MS/MS
after Trypsin Digestion and Database Searching
sample
storage
recovery (%)
m/z
origin
tuna in oil
apple puree
ale
13 weeks at 50 °C
13 weeks at 50 °C
2 months at 40 °C
2 months at 40 °C
2 months at 40 °C
9
13
46
62
36
521
676
453
465
440
621
653
988
669
731
768
peptide 198-204 þ BADGE (2þ)
peptide 413-420 þ BADGE (2þ)
peptide 310-318 þ BADGE (3þ)
peptide 460-468 þ BADGE (3þ)
peptide 298-309 þ BADGE (4þ)
peptide 139-151 þ BADGE (3þ)
peptide 118-130 þ BADGE (3þ)
peptide 184-197 þ BADGE (2þ)
peptide 469-482 þ BADGE (3þ)
peptide 529-544 þ BADGE (3þ)
peptide 139-155 þ BADGE (3þ)
lager
stout
Table 4. BADGE Adducts of Tryptic Peptides of Insulin Found by LC-MS/MS
after Trypsin Digestion and Database Searching
m/z
origin
708
943
peptide 25-47 (β-chain) þ BADGE (4þ)
peptide 25-47 (β-chain) þ BADGE (3þ)
peptide 25-47 (β-chain) þ BADGE (2þ)
peptide 85-105 (R-chain) þ BADGE (3þ)
peptide 85-105 (R-chain) þ BADGE (2þ)
peptide 85-105 (R-chain) þ 2BADGE (2þ)
BADGE, that is, BADGE 2Cys and BADGE H2O Cys (Table 2).
1415
894
3
3
3
For both insulin and BSA both reaction products were observed.
No other reaction products of BADGE and amino acids were
found. These results confirm that BADGE reacts with proteins and
that BADGE reacts mainly with the cysteine residues in proteins.
Finally, isolated proteins from tuna after reaction with
BADGE were digested using trypsin and Pronase followed by
LC-MS(/MS) analysis. Database searching of the tryptic peptides
did not show any reaction products. This is likely due to the low
abundance of these peptides as a result of the large difference
between the amount of protein in tuna, that is, 26 wt % or 260 g/
kg, and the amount of BADGE spiked to tuna, that is, 20 mg
(or 0.02 g)/kg. However, in the Pronase digest of tuna the reaction
product BADGE H2O Cys was detected, showing that BADGE
1340
1510
1348
1519
peptide 85-105 (R-chain) þ BADGE H2O (2þ)
3
3
peptide 85-105 (R-chain) þ BADGE H2O þ BADGE (2þ)
find any reaction products of BADGE with carbohydrates in
canned peaches. Further analysis using, for example, MSn,
together with detailed knowledge of the composition of the
different foods would be necessary for further identification.
Although the results in Table 3 show that a significant improve-
ment of the recovery has been achieved with the approach described
here, still not all of the BADGE was accounted for, especially for
tuna and apple puree. The most likely explanation for this is the fact
that so far only low molecular weight compounds are taken into
account. Apple puree consists mainly of carbohydrates, including
polysaccharides, and fibres. These high molecular weight com-
pounds cannot be detected by the methods applied here. It is likely
that BADGE can react with these compounds because of its
reactivity toward glucose. Furthermore, tuna consists largely of
proteins; it has been demonstrated that BADGE reacts with amino
acids, and it is likely that BADGE will also react with proteins,
which has been described recently in the literature (10).
Reactivity of BADGE with Proteins. To verify whether
BADGE reacts with proteins, both model studies and experi-
ments with food samples were carried out. Reaction of BADGE
with BSA and insulin was carried out using a proteomics
approach, that is, digestion using trypsin followed by LC-MS/
MS analysis and Mascot database searching. During database
searching, BADGE was added asa post-translation modification.
For insulin and BSA, tryptic peptides were found to which
BADGE was attached (Tables 4 and 5). These peptides were
confirmed by extracting the corresponding ions in the LC-MS
measurements. For insulin, BADGE reacted with peptide 25-47
(β-chain) and peptide 85-105 (R-chain). The latter peptide also
showed a reaction product with two BADGE molecules. From
the experiments with amino acids, it was observed that cysteine
showed the highest reactivity toward BADGE. The insulin
peptides that reacted with BADGE both contain cysteine resi-
dues, that is, two for peptide 25-47 (β-chain) and four for peptide
85-105 (R-chain). For BSA 11 BADGE adducts of tryptic
peptides were observed (Table 5). All of these tryptic peptides
contained at least one cysteine residue. However, it was not
possible from those experiments to determine exactly which
amino acids in the tryptic peptides were attached to BADGE.
To that purpose digestion with Pronase was carried out, which
usually results in single amino acids and small peptides. The
resulting Pronase digests of proteins and BADGE were searched
specifically for the two main reaction products of cysteine and
3
3
has reacted with proteins in tuna. If this approach can be made
quantitative, the mass balance for BADGE could be improved
further by taking into account reaction products with high
molecular weight compounds such as proteins.
All of these findings have relevance in the future of migration
testing for materials and articles intended for contact with foods
and sold in the European Union. Although the example studied
here involved the reaction products of the epoxy coating ingre-
dient BADGE, reaction products may be formed with food for
any reactive migrant. Although the unidentified reaction pro-
ducts are likely to be high molecular weight products of >1000
Da and therefore of no or lesser toxicological relevance, they may
be broken down into smaller fragments in the stomach. These
smaller fragments may be toxicologically important if they are
sufficiently small to be absorbed from the gastrointestinal tract.
Therefore, knowledge of the reactivity and fate of chemical
migrants in food is important, and this work demonstrates some
of the analytical tools available to study this aspect of food safety.
ABBREVIATIONS USED
BADGE, bisphenol A diglycidyl ether; LC, liquid chromato-
graphy; FLD, fluorescence detection; MS, mass spectrometry;
FT, Fourier transform; FA, formic acid; TFA, trifluoroacetic
acid; ESI, electrospray ionization; BSA, bovine serum albumin;
BFDGE, bisphenol F diglycidyl ether; Cys, cysteine; Met,
methionine; Lys, lysine; Tyr, tyrosine; His, histidine.
ACKNOWLEDGMENT
We thank H. J. Heinz for preparation of the canned food samples.
LITERATURE CITED
(1) Commission Directive 2002/16/EEC of 20 February 2002 on the use
of certain epoxy derivatives in materials and articles intended to
come into contact with foodstuffs (BADGE/BFDGE/NOGE). Off.
J. Eur. Communities 2002, L51, 27-31.

4882 J. Agric. Food Chem., Vol. 58, No. 8, 2010
Coulier et al.
(2) Commission Regulation 1895/2005 of 18 November 2005 on the
restriction of use of certain epoxy derivatives in materials and articles
intended to come into contact with food. Off. J. Eur. Communities
2005, L302, 28-32.
bisphenol A, bisphenol A diglycidyl ether and its derivatives in canned
foods by reversed-phase liquid chromatography. J. Chromatogr., A
2006, 1129, 145–148.
(9) Richard, N.; Biedermann, M.; Grob, K. Reaction of bisphenol-
A-diglycidiyl ether (BADGE) from can coatings with food com-
ponents. Mitt. Geb. Lebensmittelunters. Hyg. 1999, 90, 532–545.
(10) Petersen, H.; Biereichel, A.; Burseg, K.; Simat, T. J.; Steinhart, H.
Bisphenol A diglycidyl ether (BADGE) migrating from packaging
materials ’disappears’ in food: reaction with food components. Food
Addit. Contam. 2008, 25, 911–920.
(3) Commission Directive 97/48/EC of 29 July 1997 amending for the
second time Council Directive 82/711/EEC laying down the basic
rules necessary for testing migration of the constituents of plastic
materials and articles intended to come into contact with foodstuffs.
Off. J. Eur. Communities 1997, L222, 1-10.
(4) Minguia-Lopez, E. M.; Peralta, E.; Gonzalez-Leon, A.; Vargas-
Requena, V.; Soto-Valdez, H. Migration of bisphenol A (BPA) from
(11) Petersen, H.; Simat, T. J.; Schaefer, A.; Buckow, C. A.; Biereichel,
A.; Steinhart, H. Monomere von epoxidharzen als lebensmittelkont-
aminanten - die problematik der entwicklung einer validierten
analytik und ihre anwendung. Lebensmittelchemie 2000, 54, 127–129.
(12) Simat, T. J.; Petersen, H.; Biereichel, A.; Schaefer, A.; Steinhart, H.
Migration of epoxy monomers in canned foods and their reaction
with food components. Czech J. Food Sci. 2000, 18, 55–57.
(13) Biedermann, M.; Grob, K. Food contaminants from epoxy resins
and organosols used as can coatings: analysis by gradient NPLC.
Food Addit. Contam. 1998, 15, 609–618.
~
epoxy can coatings to jalapeno peppers and an acid food simulant.
J. Agric. Food Chem. 2002, 50, 7299–7302.
(5) Minguia-Lopez, E. M.; Soto-Valdez, H. Effect of heat processing
and storage time on migration of bisphenol A (BPA) and bis-
phenol A-diglycidyl ether (BADGE) to aqueous food simulant from
Mexican can coatings. J. Agric. Food Chem. 2001, 49, 3666–3671.
(6) Nerin, C.; Philo, M. R.; Salafranca, J.; Castle, L. Determination of
bisphenol-type contaminants from food packaging materials in
aqueous foods by solid-phase microextraction-high-performance
liquid chromatography. J. Chromatogr., A 2002, 963, 375–380.
(7) Leepipatpiboon, N.; Sae-Khow, O.; Jayanta, S. Simultaneous de-
termination of bisphenol-A-diglycidyl ether, bisphenol-F-diglycidyl
ether, and their derivatives in oil-in-water and aqueous-based canned
foods by high-performance liquid chromatography with fluore-
scence detection. J. Chromatogr., A 2005, 1073, 331–339.
(14) Rauter, W.; Lintschinger, J.; Bisphenol, F. Diglycidyl ether
(BFDGE) derivatives in canned food: synthesis and analysis. Dtsch.
Lebensm.-Rundsch. 2000, 96, 417–422.
Received for review November 30, 2009. Revised manuscript received
March 15, 2010. Accepted March 15, 2010. This work was funded by the
U.K. Food Standard Agency (A03050).
(8) Sun, C.; Leong, L. P.; Barlow, P. J.; Chan, S. H.; Bloodworth, B. C.
Single laboratory validation of a method for the determination of