Food Hydrocolloids 46 (2015) 233e243

Contents lists available at ScienceDirect

Food Hydrocolloids
journal homepage: www.elsevier.com/locate/foodhyd

Effect of globular pea proteins fractionation on their heat-induced

aggregation and acid cold-set gelation
Jean-Luc Mession*, Mohammed Lazhar Chihi, Nicolas Sok, Re
mi Saurel
Agrosup Dijon, UMR PAM 02.102, Equipe PAPC (Proc
ed
es Alimentaires et, PhysicoChimie), 1 Esplanade Erasme, 21000 Dijon, France

a r t i c l e i n f o a b s t r a c t

Article history: This report focuses on cold-set gelation of pea (Pisum sativum L.) proteins and related globulin fractions,
Received 18 April 2014 namely Vicilin 7S and Legumin 11S. Protein thermal denaturation and aggregation were investigated
Received in revised form using differential scanning calorimetry (DSC) sodium dodecyl sulfate polyacrylamide gel electrophoresis
17 November 2014
(SDS-PAGE) analysis, and sulfhydryl (S
)/disulfide (SeS) bond assessment. While the denaturation
Accepted 24 November 2014
temperature (Td) of pea proteins increased from about 69 to 77  C with increasing legumin content, the
Available online 15 December 2014
disulfide-linked acidic (a) and basic (b) legumin subunits (56e58 kDa) denatured and aggregated in a
temperature range of 75e85  C. Dissociation of legumin oligomers and their rearrangements via hy-
Keywords:
Legumin
drophobic interactions and sulfhydryl/disulfide bonds exchange reactions would occur concomitantly
Vicilin during the heat-treatment, giving rise mainly to high-molecular weight aggregates of random structure.
Thermal denaturation and aggregation However thermal denaturation and aggregation of vicilin molecules would involve solely non-covalent
Sulfhydryl/disulfide bond exchange interactions. Then glucono-d-lactone (GDL) acid-induced gelation of protein thermal aggregates was
reactions evaluated by means of G0 and G00 moduli. The preheated mixed pea globulins and vicilin-enriched
GDL-induced cold-set gelation samples gave rise to increased final moduli values of the acid gels, while legumin-enriched samples
displayed low gelling properties. Improving functional properties of pea proteins would help to promote
their application in plant-based gelled products such as tofu, alternatively to their soy counterparts.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction Soybean represent two-thirds of the world plant protein con-

Demographic change in combination with rapid urbanization
and economical growth tend to increase dramatically worldwide
pressure on land, water and natural resources. (Lundqvist et al.,
2007). In spite of the wide inequalities in access to water and
food, increases in income in most of developing countries are
associated with modifications in food consumption, with a marked
preference for animal-based food to the detriment of vegetarian
products. However, given the heavy ecological debt of meat pro-
duction systems, the need to develop alternative protein sources of
reduced bioenergy and water demands for human diet is of primary
interest. Thus one crucial challenge for food researchers would be
to increase the awareness on the utilization of sustainable plant
proteins, displaying both satisfying nutritional value and func-
tionality (Boye, Zare, & Pletch, 2010). The development of novel and
attractive plant protein-based foodstuffs would ultimately
encourage the partial replacement of ingredients provided by ani-
mal production (meat, eggs, milk).
* Corresponding author. Tel.: þ33 380 774 051; fax: þ33 380 774 047.
E-mail address: jean-luc.mession@laposte.net (J.-L. Mession).
sumption (USDA, 2014). Despite its widespread utilization in many
“vegetarian” meat-like products, the dependence on soy importa-
tions for many countries is in discordance with improving effi-
ciencies of food production (O'Kane, Happe, Vereijken, Gruppen, &
van Boekel, 2004a, 2004b, 2004c). Mainly used as protein source in
animal feed, pea (Pisum sativum L.) proteins could represent an
alternative to soybean as functional ingredient in human food
(Marcone, Kakuda, & Yada, 1998a).
Dry pea seeds contain about 20e25% crude protein, among
them 70% are storage globulins (Boye et al., 2010). Salt-soluble
globulins are composed of two fractions, namely legumin 11S and
vicilin/convicilin 7S (Tzitzikas, Vincken, de Groot, Gruppen, &
Visser, 2006). These are constituted of heterogeneous subunits
assembled into high molecular weight (Mw) oligomers. Pea legu-
min 11S is hexameric (330e410 kDa), and its subunits (mainly
60e65 kDa) dissociate into acidic a (40 kDa) and basic b (20 kDa)
polypeptides by disulfide bond reduction (O'Kane et al., 2004c).
Vicilin and convicilin 7S are trimeric (150 kDa and 180e210 kDa,
respectively) (Tzitzikas et al., 2006). Vicilin is a combination of
heterogeneous polypeptides, since the major subunits (48e52 kDa)
can be cleaved in vivo into low-Mw fragments (12e16, 20, 25e30

http://dx.doi.org/10.1016/j.foodhyd.2014.11.025
0268-005X/© 2014 Elsevier Ltd. All rights reserved.
234 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
and 30e36 kDa) (O'Kane et al., 2004a, 2004b). The convicilin sub-
units (70 kDa) display about 80% amino-acids sequence homology
with the uncleaved vicilin subunits, though are distinguishable by
their highly charged N-terminal extension region close to the C-
terminus, and also the absence of in vivo cleavage (Tzitzikas et al.,
2006).
Focusing on one essential functional property, gelation of
globular proteins is based on their ability to form three-
dimensional network (Bryant & McClements, 1998). To induce
proteineprotein interaction, a denaturing thermal process is usu-
ally applied. This leads to protein unfolding, exposure of initially
buried reactive residues and subsequent aggregation caused by
different molecular interactions, possibly involving non-covalent
and/or covalent bonds. Many recent reports pointed out the
lower textural features of pea protein heat-set gels than those ob-
tained with their soy counterparts, though processed under the
same conditions (O'Kane et al., 2004c; O'Kane, Vereijken, Gruppen,
& van Boekel, 2005; Shand, Ya, Pietrasik, & Wanasundara, 2007;
Sun & Arntfield, 2010, 2011, 2012). The type and composition of
the pea proteins used and the different procedures of each study
affected gelation properties. As reported by many authors, the
commercial pea protein isolates utilized exhibited poor gelling
properties, since proteins were extensively denatured during their
large-scale production (Shand et al., 2007). Besides, pea proteins
heat-induced gelation would require concentrated pea protein
suspensions in the presence of high salt concentrations (Sun &
Arntfield, 2010, 2011, 2012). Moreover, slow heating/cooling rates
applied would be required for pea protein molecules to form gel
network of enhanced elasticity (Sun & Arntfield, 2011). Neverthe-
less, Shand et al. (2007) indicated that the number of cross-links
within the gelled network increased with the availability of solu-
ble pea protein aggregates provided by heat-treatment.
From these observations, pea proteins cold-set gelation could
represent an alternative route (Bryant & McClements, 1998). This
procedure would enable better control of soluble protein aggre-
gation; indeed heat-denaturation and gelation are divided into two
steps. First, a low-concentrated protein solution (<10 wt%) at a pH
far from its isoelectric point and in the absence of salt is heated to
produce soluble aggregates. After cooling, gelation of the thermal
aggregates is carried out by lowering electrostatic repulsions,
allowing them to assemble into structured network. The second
step could be achieved by adding an acidifying agent, glucono-d-
lactone (GDL). Acid-induced cold gelation is well documented for
whey proteins (Alting, Hamer, de Kruif, & Visscher, 2003). In
contrast, no data was available concerning pea proteins.
In a previous work, heat-treatment of a pea globulins solution
resulted in high-Mw and soluble thermal aggregates (Mession, Sok,
Assifaoui, & Saurel, 2013). In this regard, pea proteins extraction by
an ultrafiltration/diafiltration procedure in addition with careful
control of heat-treatment would promote the usefulness of aggre-
gates as “building blocks” for cold-set gels. Hence, the present study
aimed at investigating the thermal denaturation and aggregation of
the different pea globulin fractions (mixed globular pea proteins:
PP, legumin: Leg fraction and vicilin: Vic fraction), in terms of
protein thermal stability and molecular interactions involved.
Thereafter cold-gelation properties upon acidification of the ther-
mal aggregates were compared according to the globulin fraction.
2. Material and methods
2.1. Materials
Pea proteins were extracted from smooth yellow peas (P. sat-
ivum L.), supplied by Roquette SA (Lestrem, France). Dehulled and
ground peas contained 4.1 wt% total nitrogen (TN) on a dry basis (d.
b.). The GDL powder was obtained from Prolabo (99.5% Fontenay-
sous-Bois, France).
All other reagents and chemicals purchased from Sigma-Aldrich
(St-Quentin Fallavier, France) were of analytical grade.
2.2. Methods
2.2.1. PP extraction and purification
Mixed globular pea proteins (PP) isolate was prepared from the
defatted pea flour using a salt-extraction at pH 8, followed by
tangential ultrafiltration/discontinuous diafiltration with deionized
water (UF/DF), using a polyethylenesulfone membrane of 100 kDa,
as previously described (Mession, Assifaoui, Lafarge, Saurel, &
Cayot, 2012). During the UF step, a volume concentration ratio of
3 (VCR, mL/mL) was applied. The further discontinuous DF step
consisted of a re-VCR of 3 (Taherian et al., 2011). Separation of the
globulin fractions Vic and Leg was achieved by a chromatography
column, pre-packed with DEAE Sepharose® Fast Flow anion-
exchanger (150 mL, 15 cm bed height, GE Healthcare Amersham
Biosciences Corp., Uppsala, Sweden). The pea protein concentrate
obtained by UF as described above underwent the discontinuous DF
step with 0.1 M Na2HPO4 e citric acid buffer, pH 7. Elution was
performed at 20  C using a stepwise gradient of NaCl (0, 0.06, 0.1,
0.25 and 0.5 M) in the phosphate-citrate buffer. Each eluate
(150 mL) was monitored at 280 nm and analyzed by SDS-PAGE for
purity. Polypeptide bands at 70, 50, 36e30 and lower than 14.2 kDa
were attributed to vicilin/convicilin subunits, while the large band
at 56e58 kDa belonged to legumin ab subunits (Mession et al.,
2012; O'Kane et al., 2004a, 2004b, 2004c). Eluates of identical
electrophoretic pattern were pooled together, and underwent for
each an UF (VCR of 4) and discontinuous DF (re-VCR of 3). At the
end, dried protein samples were obtained by freeze-drying, sealed
and stored at 4  C until further use. These dry samples contained
mixed globular pea proteins, enriched-fractions vicilin/convicilin
and legumin and were named PP, Vic and Leg, respectively.
2.2.2. Chemical analyses
Total moisture and ash contents were evaluated according to
AOAC (1990) procedures. Total nitrogen (TN) and non-protein ni-
trogen (NPN) were determined according to EN ISO 20483:2006
and Chavan, McKenzie, and Shahidi (2001) methods, respectively.
This allowed calculation of protein content, using a (protein)
nitrogen-to-protein conversion factor of 6.25 (Marcone et al.,
1998a; Shand et al., 2007).
2.2.3. Preparation of heat-induced aggregates
The freeze-dried PP, Vic and Leg samples were suspended in
deionized water, stirred at room temperature for at least 2 h, and
then centrifuged (12,000 g, 25 min, 4  C) to obtain protein super-
natants. To ensure that protein samples were at low ionic strength
prior to heat-treatment, the supernatants were extensively dia-
lyzed against a 10 mM Na2HPO4 buffer at pH 7.2, (ratio protein
solution-to-buffer of 1:10, 4  C, 24 h), and again centrifuged
(12,000 g, 25 min, 4  C). The PP, Vic and Leg stock solutions at 4.2,
6.2 and 3.7 wt% protein concentration, respectively, were diluted
with the phosphate buffer used for dialysis to obtain various pro-
tein concentrations: 1, 2, 4 wt% for PP and Vic samples and 1, 2,
3.5 wt% for Leg samples. Protein samples (5 mL) were poured in PX/
1636/04 MP tubes, hermetically sealed (SciLabware Ltd, Stafford-
shire, U.K.). They were placed in a water bath that was preheated at
40  C, then heated at 1  C/min from 40 to 85  C, and incubated at
85  C for 60 min. Heated tubes were subsequently cooled in ice for
30 min and stored at 4  C overnight until further use. Some protein
samples were also heated at various temperatures (70e85  C) and
times (0e60 min) for SDS-PAGE analysis (Section 2.2.5).
 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
Additionally, 20 mM N-ethylmaleimide (NEM, powder) was dis-
solved by stirring at room temperature in Leg samples especially,
and then heated as described above.
2.2.4. Differential scanning calorimetry (DSC)
Thermal parameters of the PP, Vic and Leg protein extracts in
solution were assessed with a MicroDSC III calorimeter (Setaram,
Caluire, France). An accurate weight (500 mg) of protein stock so-
lution diluted to 3 wt% protein was hermetically sealed in a 1 cm3
stainless steel vessel. The sample was heated from 25 to 100  C with
a heating rate of 0.5  C/min, using the 10 mM phosphate buffer in
the reference vessel. Onset temperature (Tonset), temperature of
denaturation (Td) and enthalpy of denaturation (DHd) were
computed from the thermograms by the thermal analysis SETSOFT
2000 software (Setaram). In addition, protein samples were heated
in the presence of 20 mM NEM.
2.2.5. SDS-PAGE
Non reducing (NR) or reducing (R)-SDS-PAGE was performed on a
discontinuous buffered system, using a triseHCl polyacrylamide
(C ¼ 2.7 wt%) stacking gel (T ¼ 4%, wt/v, pH 6.8) and running gel
(T ¼ 10%, wt/v, pH 8.9) in the presence of SDS (0.1%, wt/v) (Laemli,
1970). The dimensions of the gel were 14  16 cm. Unheated and
preheated samples prepared in Section 2.2.3 were centrifuged
(12,000 g, 25 min, 4  C). Supernatants (soluble fraction) were diluted
with deionized water on the basis of 1 wt% protein concentration for
the unheated sample. When noticed, insoluble protein sediment in
heated samples was collected and mixed at room temperature with a
denaturant solution (ratio 1:20, w/w) containing 2% SDS (wt/v), 6 M
urea and 100 mM triseHCl, pH 8.5, till complete solubilization. Pro-
tein samples were then mixed (ratio 1:1) with the sample buffer
62.5 mM triseHCl, pH 6.8, 10% glycerol (wt/v), 0.005% bromophenol
blue (wt/v) and 0.2% (wt/v) SDS, without (NR) or with (R) 1% (wt/v)
dithiothreitol (DTT). For R-SDS-PAGE, samples were heated for
10 min in boiling water. Twenty microliter of sample (5 mgprotein/mL)
was loaded in each well. Wide-range Mw standards (S8445, Sigma)
were deposited in a separated lane to identify polypeptides according
to their apparent Mw. Fixed proteins were stained with Coomasie
Blue R-250 (0.125%, wt/v) in 20% (v/v) ethanol, and destaining was
performed using 5% (v/v) acetic acid and 20% (v/v) ethanol.
The destained gels (colorless background) were scanned with a
ChemiDoc XRS þ System (Bio-Rad Lab., Hercules, CA, USA), and the
density of individual polypeptide bands in a lane was expressed as
series of staining peaks as a function of the running distance from
the border (distance 0) separating the stacking and the running gel,
using ImageLab (v. 3) software. The areas (intensities) under the
peaks were integrated and summed. This allowed calculation of
total lane intensity and thus the relative proportion of each poly-
peptide constituting the protein sample.
Upon heat-treatment of a protein sample, sulfhydryl-containing
polypeptides such as 11S globulins could denature and form co-
valent aggregates via disulfide linkages (Choi, Mine, & Ma, 2006).
Under NR conditions, the amount of these polypeptides of interest
could be quantified by densitometry, since their related band area
would decrease upon heat-treatment. On the same gel, a range of
protein concentrations (0.5e5 mgprotein/mL, i.e. 10e100 mg protein)
of the unheated sample (reference) and related samples collected
during heat-treatment (Section 2.2.3) were applied in separated
lanes. Within this protein amount loading range for sample appli-
cation, it was considered that all the polypeptides (<200 kDa) un-
der R conditions could migrate in the running gel, and total lane
intensity was correlated to the initial protein amount loaded in the
well. Because staining may vary from gel to gel, the ratio of total
lane intensity I under NR and R conditions was calculated for each
protein concentration of the unheated sample applied for
235
calibration. The INR/IR ratio served as correction factor (1),
enabling the apparent amount (concentration) of polypeptides
migrating in the running gel under NR condition. Calibration curves
were then established, plotting the band area of a polypeptide of
interest as a function of its apparent concentration.
2.2.6. Free sulfhydryl content determination
The free sulfhydryl (S
free) contents of pea protein samples were
evaluated using 5,50 -dithio-bis-2-nitrobenzoic acid (DTNB), accord-
ing to Ellman (1959). The reagents used were DTNB (1 mM) and
guanidium hydrochloride (GuHCl, 6 M) solutions, previously dis-
solved in a 0.1 M phosphate buffer, pH 7.5. Unheated and heated
(85  C/60 min) protein samples were extensively dialyzed at 4  C
against the same 0.1 M phosphate buffer, pH 7.5 (ratio protein
solution-to-buffer 1:20). The reaction mixture was prepared by
mixing 650 mL of the protein sample containing either 9 or
18 mgprotein/mL (depending on initial protein concentration) with
750 mL GuHCl and 100 mL DTNB solutions, stirred vigorously and kept
in dark for 10 min at 25  C. Absorbance was read at 412 nm, using a
molar extinction coefficient 3 for DTNB of 12,800 M
1/cm, as calcu-
lated from calibration curves with N-acetyl-cysteine in the range of
0e60 mM. The free S
content was expressed as mM S
free/gprotein.
2.2.7. Gelation experiments
Acid-induced cold gelation of thermally-aggregated protein
samples (85  C/60 min) was performed at 20  C using GDL. The GDL
powder was added to preheated protein samples, then stirred for
1 min (Table 1). Depending on protein initial concentration and/or
composition, the GDL concentration was adjusted in the range of
0.2e0.75 wt% to reach a pH of 4.5 after 4 h incubation. The pH was
measured at regular intervals with a pH meter model C561 (Con-
sort, Turnhout, Belgium), till pH reached its steady-state value.
Gel formation under isothermal conditions (20  C) was moni-
tored by measurements of the elastic (G0 ) and loss (G00 ) modulus of
the protein samples under acidification at 20  C, using a controlled-
stress rheometer (SR-5 Rheometric Scientific e Piscataway, NJ,
USA), managed by the RSI Orchestrator software (V.6.5.8). Samples
(1 mL) were placed into the parallel plateeplate geometry (25 mm
diameter, gap set at 1 mm) immediately after dispersion of GDL,
and edges of the sample were covered with mineral oil to prevent
water evaporation. Both moduli were measured as a function of
time at an oscillation frequency of 0.5 rad/s and oscillatory stress of
0.5 Pa within the linear viscoelastic region, as previously deter-
mined by a stress-sweep test (0.3e10 Pa) on gels. The gel point Gp
was considered as the time/corresponding pH value when the G0
and G00 cross-over occurred (tan d ¼ G00 /G0 ¼ 1). Gelation was
considered as achieved when each modulus plateaued around a
stable value (after about 4 h of acidification, pH in the range of
4.5e4.3). Then a frequency-sweep test (from 0.06 to 6 rad/s) was
Table 1
Composition of pre-heated protein samples prepared at the concentrations indi-
cated, for cold gelation experiments in the presence of GDL at 20  C.
Sample Protein (wt%) GDL/protein weight ratio (w/w)
PP 1 0.22
PP 2 0.21
PP 4 0.19
Vic 1 0.18
Vic 2 0.17
Vic 4 0.17
Leg 1 0.16
Leg 2 0.15
Leg 3.5 0.13
236 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
conducted at fixed stress within the linear viscoelastic domain, to
determine frequency-dependence of both moduli.
2.2.8. Statistical analysis
The presented results were obtained from two repetitions of
complete sample preparation and measurements for each prepa-
ration were at least duplicated. Analysis of variance (one-way
ANOVA) and Tukey's procedure were carried out at the p < 0.05 level
to evidence significant differences between mean values, using
Statistica software (version 7.1, StatSoft Inc, Maisons-Alfort, France).
3. Results and discussion
3.1. Protein extracts characterization
The fat content in the pea flour was 1.3 wt% (d. b.). Regarding TN
content in the defatted pea flour, about 83% was recovered in the
protein crude extract (pH 8), while 22% consisted of NPN (Mession
et al., 2012). Thus the salt-extractable protein content in the
defatted pea flour was 18.1 wt% (d. b.), in agreement with data
reported by Boye et al. (2010). After the UF/DF procedure, protein
yield from the crude extract was around 75%, while protein content
in the diafiltrated pea proteins concentrate represented 85e90% of
total solids content. Hence, non-protein material was efficiently
removed by UF/DF.
By ion-exchange chromatography applied on the diafiltrated
pea proteins concentrate, vicilin and legumin proteins eluted at
0.1 M and 0.25 M NaCl in the phosphate-citrate buffer, respectively
 & Gue
(Larre guen, 1986). As reported by these authors, it was
observed that non-globulin proteins (mainly albumins) and non-
protein components eluted as broad peaks of strong UV-
absorption at 0 and 0.5 M NaCl, respectively (elution profile not
shown). Protein yields for both Vic and Leg-enriched fractions were
in the range of 8e10% of the starting PP content loaded in the
column. Protein contents in the freeze-dried PP, Vic and Leg-
enriched fraction samples were 85.8 wt%, 54.2 wt% and 56.5 wt%
(d. b.), respectively. Salt contents were 4 wt% for PP and around
30e35 wt% for both enriched-fractions (d. b.). Despite the UF/DF
procedure performed on either Vic or Leg-enriched fraction, the
salts brought during the purification step remained in large
amounts in their related freeze-dried samples (concentrates).
Thereby the dialysis step following the dissolution of freeze-dried
samples in the 10 mM phosphate buffer (pH 7.2) was a cautious
approach (Section 2.2.3). Ultimately total protein accounted for
more than 90% of total solid in protein stock solutions.
3.2. Differential scanning calorimetry (DSC)
Thermal properties of PP, Leg and Vic samples at pH 7.2 were
evaluated using DSC, firstly in the absence of NEM additive
(Table 2). All the thermograms (data not shown) displayed one
Table 2
Thermal parameters of pea protein extracts (3 wt%, heated at 0.5  C/min) in 10 mM Na2HPO4 buffer, pH 7.2, in the absence (
NEM) and in the presence (þNEM) of 20 mM N-
ethylmaleimide.
Sample Tonset ( C)a

NEM þNEM
PP 63.3 ± 2.6bA 62.2 ± 1.4bA
Leg 69.3 ± 1.3aA 70.6 ± 0.5aA
Vic 64.9 ± 2.5bA 63.0 ± 2.7bA
Column values followed by the same lowercase letter (a and b) are not significantly different (p < 0.05).
Effect of NEM addition on each thermal parameter of a same pea protein extract: row values followed by the same capital letter (A and B) are not significantly different
(p < 0.05).
a
Mean ± SD.
broad endothermic peak that was attributed to thermal denatur-
ation of 7S and/or 11S globulin fractions. The Tonset value is the
temperature from which unfolding of the globular protein structure
is initiated. This value was of about 64  C for both PP and Vic
samples, 5  C lower than that measure for Leg. The Td value is the
peak thermal denaturation temperature; a high value indicates
usually a high thermal stability for globular proteins (Choi & Ma,
2005). Both PP and Leg samples displayed a Td value in a narrow
range of 75e77  C, at least 6  C higher than that measured for Vic.
The DHd value is the content of ordered structure of a protein, or
can be related to the proportion of non-denatured proteins before
heat-treatment (Sun & Arntfield, 2010). There was no significant
difference between the protein samples investigated, despite mo-
lecular structure differences of the extracted pea globulins.
Both Td and DHd values could be indicative of the preservation of
the “native” globulin structure during the extraction procedure. To
compare with previous data, the PP sample had a Td value of about
10  C lower than that reported for pea globulins prepared by Shand
et al. (2007) and Sun and Arntfield (2010). The PP sample had a DHd
value close to that measured by Shand et al. (2007). The Td value
measured for the Vic sample was in agreement with that found in
similar conditions by O'Kane et al. (2004a) for pea vicilin fractions,
however the Leg sample had lower thermostability of about 7  C
than that reported by Marcone, Kakuda, and Yada (1998b). Mean-
while in the present study, higher pea proteins thermostability
could be indicative of higher legumin molecules content in the
order Vic < PP < Leg samples (Table 3), since legumin hexamers
were of higher conformational compactness e thus harder to
thermally-unfold e than that for vicilin trimers (Marcone et al.,
1998b). Additionally, the presence of disulfide bonds within legu-
min subunits Lab would contribute to the thermal stability of the
protein, as reported for its buckwheat legumin counterpart (Choi
et al., 2006). However discrepancies regarding thermal parame-
ters of pea globulins were particularly related to the pea cultivar
and the legumin-to-vicilin ratio; while vicilin proteins were prac-
tically devoid of sulfur-amino acids, the number of cysteine residue
per legumin subunit could vary between 2 and 7 (O'Kane et al.,
2004c, 2005; Sun & Arntfield, 2012). Moreover, extrinsic factors
enhancing protein compact packing and/or proteineprotein in-
teractions could be mentionned in particular the stabilizing effect
of NaCl (at pH higher than 6) towards globulin quaternary structure
and/or an increase of the level of energy required for protein
unfolding with an higher heating rate applied (Mession et al., 2013;
Sun & Arntfield, 2011). The different experimental conditions used
rendered difficult comparisons between the authors regarding
heat-denaturation of pea and other plant proteins.
With respect to the possible involvement of sulfhydryl/disulfide
bonds (S
/SeS) exchange reactions during pea globulins denatur-
ation and aggregation, NEM (20 mM) was added to protein samples
prior to DSC analysis (Table 2). NEM is a sulfhydryl-blocking agent
(O'Kane et al., 2004c). All thermograms showed one endothermic
Td ( C)a DHd* (J/gprotein)a

NEM þNEM
NEM þNEM
74.5 ± 1.6aA 73.7 ± 1.4aA 7.9 ± 1.5aA 6.2 ± 0.0aA
76.6 ± 0.4aA 72.1 ± 0.3aB 10.3 ± 1.4aA 7.5 ± 0.6bB
68.5 ± 1.0bA 71.1 ± 2.8aA 7.6 ± 1.1aA 8.3 ± 0.8bA
 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
peak (data not show). For PP, the overlapping thermal transitions of
pea proteins may hide small differences in their respective Td values
(Shand et al., 2007). In the presence of NEM, the Tonset value of Leg
was higher by 5  C as those for PP and Vic, while the Td values close
to 72  C were not found to differ. Concerning DHd, the PP þ NEM
sample displayed the lowest value, however in the same range than
that measured for the PP e NEM sample. By comparing each protein
sample in the presence or in the absence of NEM, it appeared that Td
and DHd values were solely affected for the Leg þ NEM sample,
decreasing of about 4.5 and 3 J/gprotein, respectively. These results
were indicative of the limited effect of NEM on PP thermal dena-
turation, whereas there was no apparent effect for Vic. Firstly, NEM
reactivity would be lowered since sulfhydryl groups were deeply
buried within legumin structure (O'Kane et al., 2004c). Secondly,
non-covalent interactions were solely involved in the stabilization
of the pea globulins oligomeric structure (Marcone et al., 1998a);
heat-induced dissociation of pea globulin subunits would be a
prerequisite to allow sulfhydryl exposure and thus protein struc-
tural changes by covalent bonding. Therefore NEM could induce
slight conformational change of legumin molecules. As NEM
inhibited disulfide bonds formation, it could be hypothesized that
non-covalent bonds, of lower energy than the former ones, were
promoted during heat-denaturation of legumin molecules thus
facilitating their unfolding and subsequent aggregation via non-
specific interactions. This assumption is supported by the very
narrow temperature range of 2  C between Tonset and Td values of
Leg þ NEM, indicative of a highly cooperative endothermic tran-
sition, i.e. occurring simultaneously at several levels of structure
(Marcone et al., 1998b). Besides, this was consistent with the lower
DHd value for Leg þ NEM than that for Leg e NEM sample, possibly
reflecting the higher contribution of exothermic events than
endothermic ones during thermal denaturation of Leg þ NEM (Choi
et al., 2006). Meanwhile, the decreasing thermal stability of
Leg þ NEM was not evidenced by O'Kane et al. (2004c) for purified
pea legumin. This may account for the less native state of the
fraction Leg in this study, probably increasing legumin molecules
sensitivity towards NEM.
By reheating one replicate of each protein sample, no thermal
transition was detected; pea protein denaturation was complete
and irreversible.
3.3. SDS-PAGE
Polypeptide composition of unheated pea protein samples was
investigated by NR- or R-SDS-PAGE (Fig. 1a: PP, lanes P and P*; 1b:
Leg, L and L*; 1c: Vic, V and V*). Electrophoretic patterns displayed
numerous polypeptide bands ranging from 90 kDa to less than
14.2 kDa. Band identification was performed on the basis of Shand
et al. (2007), and O'Kane et al. (2004a, 2004c) reports.
The polypeptides of Mw 85e90 kDa were attributed to lip-
oxygenases Lox1e2. The legumin subunit Lab (Fig. 1a and b, lanes P
and L, Mw 56e58 kDa) was characterized by its dissociation under
R conditions (þDTT) into acidic La (38e42 kDa) and basic Lb1e2
polypeptides (24 and 18 kDa, noted Lb in the following for simpli-
fication, lanes P* and L*). The band caught at the top of the stacking
gel was reported to be disulfide-bonded legumin polymers; these
are usually encountered in “native” (unheated) pea protein samples
and were dissociated under R conditions (Marcone et al., 1998a).
Additionally, some La and Lb2 polypeptides appeared not to be
disulfide-bonded in both unheated PP and Leg samples. The vicilin
polypeptides were composed of convicilin (Cv), uncleaved Vi1
subunits (52e48 kDa), and cleaved ones Vi2 (37e30 kDa) and also
Vi3 (22e14.2 kDa) (1c, lane V). A polypeptide of Mw 60 kDa was
detected (namely NV: non-vicilin) and though vicilin subunits were
not held together by disulfide bonds, the former one dissociated
237
under R conditions (Lane V*). Since any polypeptide band assign-
able to either La or Lb polypeptides was noticed in V*, there would
be no cross-contamination of the Vic fraction by Lab. Nonetheless,
remaining legumin in Vic could not be totally ruled out; the very
small increase under R conditions of staining intensity just below
the Vi1 bands would suggest the presence of a minor acidic legumin
polypeptide of slightly higher Mw (faint band around 44 kDa) than
that determined for La (Table 3). The NV band could be otherwise
assigned to disulfide-bonded albumins co-extracted with vicilin
molecules during the chromatography procedure, of Mw lower
than 14.2 kDa for the monomers under R conditions (Larre  &
Gueguen, 1986; Shewry, Napier, & Tatham, 1995).
From the electrophoretic patterns, a densitometric analysis was
performed to evaluate the relative amounts of the polypeptides
encountered in unheated samples (Table 3). For PP in NR condi-
tions, ratios of vicilins/legumin (Vi1e3/Lab) and vicilins/convicilin
(Vi1e3/Cv) were of about 2.4 and 5.7, respectively. Besides, a legumin
content close to 25% was found in most of pea cultivars (O'Kane
et al., 2005; Tzitzikas et al., 2006). The enrichment of the Leg and
Vic samples by Lab and Vi1e3, respectively, were of about 2.5 and 1.5-
fold higher than their relative amounts in PP. About 25% of the total
amount of La and Lb polypeptides (R conditions) was not disulfide-
bonded in the Leg sample under NR conditions. The increase in the
relative amount of Vi3 in Vic under R conditions compared to that
found under NR conditions was possibly attributable to traces of
dissociated basic legumin polypeptides (18e22 kDa), and/or albu-
min monomers.
Prior to electrophoresis, the heated Leg samples developed
strong turbidity at 80  C, while PP and Vic remained slightly
yellowish. By centrifugation, precipitation was evidenced for
heated Leg; it was assessed that around 85e90% of the initial
protein content in unheated Leg samples remained soluble in su-
pernatant of heated ones, while this was more than 95% for both PP
and Vic.
The NR-SDS-PAGE profiles of soluble proteins in heated PP and
Leg samples (Fig. 1aeb, from the lanes SP-75  C and SL-75  C,
respectively) exhibited components of low electrophoretic mobility
(Mw > 200 kDa), since they entered hardly in the running gel.
Additionally, a new polypeptide band at 85 kDa (noted LA) was
detected in the case of heated Leg. During heat-treatment, the Lab
band intensity decreased for both PP and Leg samples. This was
indicative of pea proteins heat-induced denaturation initiated at
70e75  C.
Under R conditions, the disappearance of the high Mw-
aggregate and LA bands was related to their dissociation, and
electrophoretic patterns of unheated and heated samples were
comparable; the heated PP and Leg samples exhibited the charac-
teristics bands attributed to La and Lb polypeptides (Fig. 1aeb, lanes
SP*-, SL*-85  C/60 min). Therefore thermal denaturation of legumin
subunits Lab would lead to the formation of non-covalent and co-
valent aggregates, with the possible involvement of intermolecular
disulfide bonding. Besides, soluble and insoluble legumin fractions
had similar patterns, except the low band resolution for the latter
one, which could be related to enhanced protein denaturation
(lanes IL- and IL*-85  C/60 min).
The occurrence of disulfide linkages in heated Leg samples was
further investigated by heating the fraction Leg with added NEM
(Fig. 1b0 ). Protein precipitation was markedly noticeable from 75  C,
due to the insolubilization of the Lab subunits (lane ILN-
85  C/0 min). Remaining soluble polypeptides were residual vicilin
Vi2 polypeptides (from the lane SLN-75  C) (36e20 kDa). Surpris-
ingly, further heat-treatment induced the breakup of intermolec-
ular disulfide linkages between La and Lb polypeptides of insoluble
legumin (lanes ILN- and ILN*-85  C/60 min). The present results
suggested restricted intramolecular S
/SeS exchange reactions
238 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243

Fig. 1. Electrophoretic patterns (SDS-PAGE, 12%) of unheated (40  C, prior to heat-treatment) PP (a, lanes P and P*), Leg (b, lanes L and L*), Leg þ 20 mM NEM (b0 , lane LN) and Vic
(c, lanes V and V*) samples in non-reducing (NR) or reducing (R) conditions (in the presence of DTT, as indicated by the asterisk“*”). A protein amount lower than or equal to 100 mg
was deposited in each lane. Under the lanes, were soluble (S) proteins after centrifugation of heated PP, Leg, Leg þ NEM and Vic samples at various temperature (70e85  C) and
incubation time at 85  C (0e60 min), while IL and ILN were insoluble protein material encountered in the Leg and Leg þ NEM samples, respectively. On the lanes, Mw M: molecular
weight markers, polypeptide band Lox: lipoxygenases 1e2, Cv: convicilin, NV: non-vicilin protein, Lab: main legumin subunit, Vi1e3: bands attributed to vicilin 7S, La and Lb1e2: acidic
and basic polypeptides constitutive of the main legumin subunit, respectively, LA and VA: disulfide-bonded protein aggregates (<200 kDa) formed in heated Leg and Vic samples,
respectively.

deep within La and Lb polypeptides that were not inhibited by NEM,
owing to low accessibility of sulfhydryl groups and/or probably
their closeness on polypeptide chains. The lower thermostability as
measured by DSC (Table 2) for the Leg þ NEM sample could arise
from easier unfolding and higher polypeptide chain flexibility in
the presence of NEM that enhanced aggregation via non-specific
interactions, whereas the reduction in DHd would be related to
the formation of insoluble aggregates (Choi & Ma, 2005). Since
NEM modified thermal denaturation of Lab, the newly-formed
intermolecular disulfide bonds between denatured legumin mol-
ecules would be required to maintain a stable conformation of the
heat-induced aggregates.
The heated Vic samples displayed from 75  C several novel
polypeptide bands noted VA (Fig. 1c, from lane SV-75  C, Mw
100e200 kDa), while intensity of the polypeptide band NV (60 kDa)
started to decrease between 80 and 85  C/0 min; thus the NV
 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243 239

Table 3
Densitometric analysis of the electrophoretic patterns (Fig. 1aec) of unheated PP (1a, lanes P and P*), Leg (1b, lanes L and L*) and Vic (1c, lanes V and V*) samples in either non-
reducing (NR) or reducing (R) conditions. Polypeptide band assignation is displayed in Fig. 1. The relative amounts of the different polypeptides were calculated from the
intensity of the band of interest relatively to the sum of total migrating polypeptide bands in the same lane of the running gel electrophoresis.

Polypeptide Mw (kDa) % PPa % Vica % Lega

NR R NR R NR R

Lox 87 7 ± 0.35 5.7 ± 0.5 2.5 ± 0.7 2.9 ± 1.5 5.4 ± 0.2 5.8 ± 0.3
Cv 66 8.7 ± 2.3 8.6 ±2 13.3 ± 0.3 12.4 ± 1.5 3.4 ± 1.1 2.2 ± 0.0
NV 60 e e 9.5 ± 0.4c 3.7 ± 0.8 e e
Lab 56e58 21.2 ± 3.2b e e e 65.1 ± 0.8b 3.1 ± 0.8
Vi1 52e48 13.8 ± 2.9 12.2 ± 1.8 24.5 ± 2.7 25.6 ± 2.4 e e
La 42 5.2 ± 1.6 16.2 ± 1.3 e e 12.2 ± 0.8 42.4 ± 1.1
Vi2 36e30 16.2 ± 0.9 17.9 ± 0.7 21.8 ± 0.4 21.6 ± 3.3 6.7 ± 1.3 8.3 ± 1.6
Lb1 24 3.2 ± 0.6 8.3 ± 1.3 e e 3.1 ± 0.8 13.0 ± 0.2
Lb2 18 5.0 ± 1.0 8.8 ± 1.6 e e 4.8 ± 0.4 25.1 ± 0.8
Vi3 22e20; 14.2 20.2 ± 1.4 21.9 ± 1.1 27.3 ± 3.0 34.3 ± 2.6 e e
a
Means ± SD.
b
The NV polypeptide band could not be quantified separately from the Lab band, owing to close running distance and thus Mw.
c
The NV band could be evaluated accurately for the Vic sample.

concentration decreased from 0.9 to 0.2 mgNV/gsolution at temper-
atures below (or equal to) 80  C and 85  C/60 min, respectively
(densitometric analysis not shown). The vicilin bands Vi1e3 were
not affected by the heat-treatment. Patterns of unheated and
heated Vic samples were found similar under R conditions (lanes V*
and SV*-85  C/60 min). Thus thermal denaturation of vicilin mole-
cules and their aggregation may involve solely non-covalent in-
teractions, whereas the formation in low amounts of disulfide-
bonded covalent aggregates would be attributable to NV and/or
sulfur-containing albumins.
A densitometric analysis was conducted on SDS-PAGE patterns
of soluble proteins to study change in protein composition during
heat-treatment of PP (1a, lanes SP-) and Leg (1b, lanes SL-) samples,
from 1 to 4 wt% and 1 wt% protein concentration, respectively. The
thermal denaturation and aggregation of Lab subunits was of
peculiar interest (Fig. 2). For gel calibration, a linear relation was
established between the Lab band intensity and its relative con-
centration in the lane according to different dilution factors used
for the unheated samples. Thus the decrease in the Lab band in-
tensity was related to its decreasing concentration by thermal
denaturation, leading to either covalent aggregation or precipita-
tion. The Lab concentration was shown to decrease drastically
during the heating ramp (Fig. 2). For all the protein samples, it was
worth noting that half of the initial Lab subunits content was
already denatured at 80  C (Fig. 3a). From 85  C/0 min, the decrease
in the Lab concentration slowed down, and plateaued at 85  C/
30 min toward values in the range of 10e20% of the initial con-
centration in the unheated sample, independently of the total
protein amount which underwent heat-treatment. Moreover,
calculating the ratio 1
INR/IR of total lane intensity I could estimate
the proportion of protein material involved into soluble disulfide-
bonded aggregates (Fig. 3b). The breakup under R conditions of
intermolecular disulfide linkages would allow the migration of
polypeptides involved into high-Mw aggregates (>200 kDa) that
could not be quantified under NR conditions in the running gel of
the electrophoresis. It was checked that INR remained lower than IR
for a same protein amount deposited (100 mg) in the lane and pre-
treated under each condition, in a same electrophoresis gel; this
was significant of the increased electrophoretic mobility and thus
higher content in migrating polypeptides of the reduced sample.
Therefore the noticeable decrease in INR with increasing tempera-
ture, in parallel to the Lab denaturation, was interpreted as the in-
crease in the relative amount of covalent aggregates in heated

Fig. 2. Evolution of the legumin concentration (Lab subunit, z56e58 kDa, mgprotein/
gsoluble fraction, left axis) during the heat-treatment applied (dotted line, right axis: 1  C/
min from 40 to 85  C, then 85  C from 0 to 60 min), as measured by densitometric
analysis of the electrophoretic patterns under NR conditions of unheated and heated
PP (1, 2 and 4 wt% initial protein concentration; Fig. 1a for 1 wt%; 2 and 4 wt% are not
shown), and fraction Leg (Fig. 1b, 1 wt% protein).
Fig. 3. Change in protein composition (left axis) during heat-treatment (dotted line,
right axis), as evaluated by densitometric analysis of the electrophoretic patterns of
unheated and heated PP (1, 2 and 4 wt%) and fraction Leg (1 wt%) samples. (a): con-
centration ratio of the Lab subunit (z56e58 kDa), relative to that in the unheated
sample (time 0, 40  C, Fig. 2). (b): calculated ratio 1
INR/IR of total lane intensity for a
same protein sample collected at given temperature and then pre-treated under NR
and R conditions, respectively.
240 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
samples; this was particularly prominent during the heating ramp
(75e85  C/0 min). For heated Vic, PP and Leg samples at 1 wt%, the
calculated relative amounts of high-Mw covalent aggregates were
around 13%, 25% and 52%. Consequently these were more promi-
nent with increasing the initial legumin content in heated pea
protein samples. Regarding the occurrence of the LA aggregates
(85 kDa) in the heated Leg sample, their concentration at 85  C/60
was practically ten-fold lower (0.4 mgLA/gsolution) than that for non-
migrating ones (>200 kDa).
From the densitometric profiles of SL and SP lanes, it should be
noted that band intensities under NR conditions of unbound La
and Lb polypeptides were not affected upon heating; there was no
significant dissociation of legumin subunits by extensive dena-
turation, thus no preferential involvement of La or Lb polypeptides
into covalent aggregation. This constituted a crucial difference
with heat-treated 11S from buckwheat or soy glycinin (Choi et al.,
2006). However, the hypothesized mechanism of pea legumin
thermal aggregation would be compared with that reported for
oat globulin 11S, despite the use of less severe heat-treatment and
lower ionic strength in the present study (Ma & Harwalkar, 1987);
with respect to Td values of PP and Leg samples (Table 2), loss of
the legumin oligomeric structure by weakening of hydrophobic
interactions and subsequent S
/SeS exchange reactions between
the Lab subunits were suggested to be initiated from 75  C.
Intermolecular disulfide bonding between the Lab subunits in the
early stages of heat-treatment would decrease the extent of their
denaturation, and consequently further molecular rearrangements
via non-specific interactions were lowered. The unachieved legu-
min unfolding was assumed to result in stiff protein aggregates,
which remained in majority soluble contrary to that was observed
in the presence of NEM (Fig. 1b0 ). By incubating at 85  C, the de-
natured legumin molecules would further aggregate covalently by
incorporating more Lab subunits, which involvement reached ul-
timately (85  C/60 min) about 80e90 % of their initial amount
(Fig. 3).
For heated PP at either 1 or 2 wt% protein concentration, Fig. 3
showed that disappearance and subsequent covalent aggregation
of the Lab subunits occurred similarly from 80  C, while these were
slightly alleviated at 4 wt%. From this, it was suggested that both
steric hindrance and electrostatic repulsions caused by vicilin and
convicilin molecules in the PP samples could decrease aggregation
of denatured legumin via non-covalent bonds, possibly preventing
macro-aggregates formation. In fact the presence of convicilin was
previously shown to reduce gelation ability of pea protein samples
and thence aggregation via non-specific interactions, especially for
protein isolates of low cysteine content (O'Kane et al., 2005).
3.4. Sulfhydryl content
Free sulfhydryl groups content (S
free, located on surface and
inside globulin structure) of unheated PP, Vic and Leg samples was
evaluated in the presence of the denaturant GuHCl (Table 3). The PP
sample displayed the highest content, three-fold higher than that
measured in both Leg and Vic samples. However legumin mole-
cules were reported to be the main sulfur-amino acids source
among pea globulins (O'Kane et al., 2005). Thus regarding the S
free
content of PP, these could originate from pea albumins 2S of high
cysteine content (Shewry et al., 1995). In turn, the S
free found in
the Vic sample could arise from remaining albumins, the presence
of the disulfide-bonded NV polypeptide, and possibly convicilin
which could contain one or two cysteine residues (O'Kane et al.,
2005). The low S
free content in Leg may be on account of the
pea cultivar from which the pea seeds originated from, and also the
involvement to a large extent of sulfhydryl groups of legumin
subunits in disulfide bridges, as has been mentioned for its 11S
buckwheat counterpart (Choi et al., 2006). Based on a mean average
Mw of 58 kDa for the Lab subunit and thus 350 kDa for the hexamer,
legumin in the Leg sample would contain z1.2 mol S
free/mol. By
comparison, O'Kane et al. (2005) reported for pea protein isolates
from various cultivars a S
free content in the range of 3e70 mmol
S
free/gprotein, strongly suggesting that the pea proteins used here
contained a few sulfur-amino acids.
The S
free content of heated and aggregated PP, Vic and Leg
samples at various concentrations was evaluated, to determine the
formation of new disulfide bridges (Table 4). For all the initial
protein concentrations investigated, heated Vic samples had the
lowest S
free content, while contents were not found to differ be-
tween heat-treated PP and Leg. By considering separately each
protein fraction, the S
free content was unaffected for PP and Vic
samples, while it increased two-fold for Leg. Such a tendency was
independent of the initial protein concentration that was pre-
heated. The relatively constant S
free content for unheated and
heated PP samples was related to the balance between breakup of
existing disulfide bridges in non-denatured proteins and the new
ones established by thermal denaturation and covalent aggregation
(Choi et al., 2006). For the fraction Vic, the low S
free content in the
unheated sample would minimize potentially disulfide bonding by
heat-treatment. Concerning heated Leg samples, the increase in
S
free content would originate from the disruption of preexisting
disulfide bridges. Therefore upon heat-treatment, the denatured
legumin subunits would rearrange into large aggregates via hy-
drophobic interactions and/or disulfide linkages, suggested to occur
concomitantly (O'Kane et al., 2005). However a few released S
free
groups would be involved in the formation of intermolecular di-
sulfide bonds, since the simultaneous exposure of hydrophobic
groups would result in non-specific interactions between unfolded
polypeptide chains (Choi & Ma, 2005). Consequently the propensity
for random aggregation of denatured legumin molecules would
decrease the accessibility of S
free groups.
In the following, the cold-set gelation procedure was performed
on the total (soluble þ insoluble) protein content of preheated
samples (no centrifugation of the preheated Leg samples), to avoid
protein loss and thus to allow direct comparisons between samples
prepared at the same initial protein concentration.
3.5. GDL-induced acid gelation of pea globulins
3.5.1. Kinetics of acidification
Cold-gelation of preheated pea protein samples (85  C/60 min)
was performed with added GDL. As extensively reported for whey
proteins, the protons released with time by GDL hydrolysis led to
the progressive reduction of electrostatic repulsion between pro-
tein aggregates. These were incorporated into growing clusters by
physical (non-covalent) interactions, which in turn could interact
Table 4
Free sulfhydryl contents (mmol S
free/gprotein) in unheated and heated pea protein
samples, prepared at various initial protein concentrations as indicated.
Sample Protein wt%
Unheateda 1a 2a 3.5a 4a
PP 8.9 ± 0.8aA 6.5 ± 0.4aA 7.4 ± 1.0aA ND b
7.0 ± 0.3aA
Leg 2.8 ± 0.4bB 5.4 ± 0.3aA 5.4 ± 0.1aA 5.3 ± 0.8A NDb
Vic 2.9 ± 0.4bA 2.3 ± 0.1bB NDb NDb 2.4 ± 0.2bB
Column values followed by the same lowercase letter (a and b) are not significantly
different (p < 0.05).
Effect of the initial protein concentration (of the same protein fraction) that un-
derwent thermal denaturation and aggregation: row values followed by the same
capital letter (A and B) are not significantly different (p < 0.05).
a
Means ± SD.
b
Not determined.
 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243 241

physically with each other to form a space-filling network (Alting Table 5

et al., 2003). Gelation properties at 20  C of pre-heated protein samples (85  C/60 min) in the
presence of GDL, prepared at various initial protein concentrations.
The amount of GDL was adjusted owing to the buffering capacity
of proteins and their initial concentrations (Table 1). Hence, the Sample Viscoelastic Protein wt%
preheated PP, Vic and Leg samples displayed comparable kinetics of properties
1a 2a 3.5a 4a
acidification profiles (Fig. 4, the preheated PP samples were given b e
PP Gp (min) 10.1 ± 2.9bA 9.2 ± 2.3bA ND 5.6 ± 1.2bB
as an example); a pH decrease of 2 units occurred in the first 2 h, Gp (pH)b 6.7 ± 0.2aA 6.6 ± 0.2aA 6.7 ± 0.1aA
followed by a slower decrease of 0.7 units within 2 h. A pH value of G0 (Pa)c 52 ± 7aC 178 ± 13aB 567 ± 74aA
about 4.5e4.3 was measured after 4 h incubation with GDL, within tan dc 0.2 ± 0.0bA 0.2 ± 0.0bA 0.21 ± 0.01bA
the pH range (4e5) of minimum solubility as measured for pea Leg Gp (min)b No geld No geld 7.7 ± 1.5 NDe
proteins (Shand et al., 2007). The steady-state pH values of around Gp (pH)b 6.6 ± 0.0
4.0e4.3 were reached after 6 h incubation with GDL. It was ensured G0 (Pa)c 200 ± 55.9
tan dc 0.2 ± 0.0
that higher amounts of GDL were inadequate, since proteins re-
solubilized at pH values at equilibrium below 3.8; this was Vic Gp (min)b 20.7 ± 2.1aA 19.5 ± 2.1aA NDe 20.0 ± 2.0aA
Gp (pH)b 6.1 ± 0.2aA 5.9 ± 0.2bA 6.0 ± 0.1bA
ascribed to the increasing solubility of pea proteins at acidic pH
G0 (Pa)c 44 ± 12aB 84 ± 12bB 619 ± 40aA
(Mession et al., 2012). Regarding concentration ratios GDL/protein tan dc 0.23 ± 0.01aA 0.24 ± 0.01aA 0.23 ± 0.01aA
(wt/wt), it appeared that the heated fraction Leg had the lowest
To compare each viscoelastic property, column values followed by the same
buffering capacity. lowercase letter (a and b) are not significantly different (p < 0.05).
To overcome the effect of GDL hydrolysis, the time scale during Effect of initial protein concentration (of the same protein fraction) that underwent
gelation experiments was converted into pH values. thermal denaturation and aggregation: row values followed by the same capital
letter (A
C) are not significantly different (p < 0.05).
a
Means ± SD.
3.5.2. Rheological measurements b
Time (and corresponding pH) to observe the gelation point Gp, which was
Acid-induced gelation at 20  C of preheated/aggregated pea defined as the last cross-over of G0 and G00 moduli (tan d ¼ 1).
protein samples (listed in Table 1) was characterized using oscil- c
Elastic modulus (G0 ) and loss factor (tan d) plateau values.
d
latory rheological measurements. At the beginning of the acidifi- The rheological test evidenced wide discrepancies in measurements between
replicates; sample instability would arise from protein coagulation (no true gel).
cation process, all the protein exhibited a predominant viscous e
Not determined.
behavior (G00 > G0 , tan d > 1), till the elastic component increased
dramatically (G0 > G00 , tan d < 1) from the gel point noted Gp
(tan d ¼ 1), indicative of the solegel transition (Sun & Arntfield,
2010). Time to observe Gp and corresponding pH (as obtained concentration of 1 wt%, with final tan d values in the same range as
from the acidification profiles, Fig. 4) were measured (Table 5). that measured at higher protein concentration (Table 5). Structural
Preheated Leg samples at either 1 or 2 wt% initial protein concen- rearrangement within the acidified sample were suggested to occur
tration were unstable during the rheological measurements; at the onset of gelation, giving rise to clusters of protein aggregates
among replicated samples, some apparently evidenced gelation, via physical interactions, and consequently an increase in G0 value.
however G0 plateaued at values below 10 Pa, while tan d remained This was noticed previously for acid gels made of preheated whey
around 0.35; this could reflect coagulation rather than gelation, proteins (Alting et al., 2003). For both PP and Vic samples at 1 wt%
since the presence of macro-aggregates of low solubility were re- concentration, the protein amount was too low to form a self-
ported to be detrimental for gel formation (Sun & Arntfield, 2010). supporting gel (G0 at plateau lower than 100 Pa, Table 5). Mean-
It appeared that gelation was triggered at higher pH values for all while at 2 wt%, the increase in the G0 plateau value was of about
the PP samples than those found for Vic samples, around 6.6 and 6, four-fold and two-fold for PP and Vic, respectively; a longer gelation
respectively, with no significant difference regarding the various time (larger pH range) for PP than that for Vic would allow more
concentrations of each preheated protein samples. cross-links between the thermal aggregates, which arranged into
Surprisingly preheated PP and fraction Vic samples could form more structured network. In contrast, the Vic sample required a
weak gels (low G0 value at apparent plateau) from a protein higher level of acidification prior to Gp, possibly attributable to the
higher amount of repulsive charges between aggregates to screen.
The increase in G0 was more marked from 2 to 4 wt% preheated
protein concentration, than that observed from 1 to 2 wt%; a higher
amount of denatured protein were incorporated in the gel network
since the number of possible cross-links would increase (Fig. 5).
From Gp, a one hundred-fold G0 increase was observed in a pH
range of 6.7e6, 6.6e5.7 and 6e5.8 for preheated PP, Leg and Vic
samples at 4, 4 and 3.5 wt% initial protein concentration, respec-
tively. Therefore network formation was particularly straightfor-
ward from pH 6 concerning the Vic fraction. Protein network
strengthened while pH decreased, as G0 increased more rapidly
than G00 . The tan d values stabilized at pH 5.7, close to 0.2 for both PP
and Leg samples, whereas it was slightly higher for Vic. At pH 4.8,
both G0 and G00 moduli stabilized. The preheated PP and Vic samples
had close G0 plateau values (Table 5). Meanwhile, the G0 plateau
value for the Leg sample was shown to vary markedly between
replicated samples; as shown by O'Kane et al. (2004c) for heat-set
gels, legumin had the propensity to assemble into heterogeneous
network of coarse structure. As the fraction Leg displayed a higher
Fig. 4. Change in pH at 20  C in the presence of GDL, of pre-heated/aggregated PP minimum protein aggregation concentration to allow cold-gelation
samples prepared at various initial concentrations (protein wt%, Table 1). than those obtained for both PP and Vic, this was indicative of the
242 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
Fig. 5. Variation of (a) storage modulus G0 and (b) tan d at 20  C as a function of
decreasing pH in the presence of GDL, for preheated/aggregated PP, fractions Vic and
Leg samples prepared initially at 4, 4 and 3.5 wt% protein concentration, respectively
(Table 1).
lower gelling abilities of preheated Leg. From Figs. 2e3 and Table 3,
it was suggested that the concomitancy of thermal unfolding and
aggregation of legumin molecules led to random arrangements
between partially-denatured proteins. Moreover, protein precipi-
tation in heated Leg samples would be indicative of macro-
aggregates formation, which was reported to be detrimental for
the formation of a structured gel network; more likely coagulum
was obtained (Sun & Arntfield, 2010).
When viscoelastic moduli of acid gels were found to be stable,
frequency sweep tests were carried out (Fig. 6). Spectra displayed a
G0 modulus higher than G00 in the whole frequency range. Due to the
relaxation of protein molecules that were not incorporated into the
gel network (viscous contribution), both moduli decreased at low
frequencies. This was typical of food protein gels, which loose-
gelled network exhibited viscoelastic properties (Alting et al.,
2003); for acid gels, the protein network involved predominantly
physical bonds between pea protein thermal aggregates, which in
turn could rearrange extensively as their structure would be mainly
held by non-covalent forces. Additionally, the tan d plateau values
of about 0.2 were significant of weak acid gels, in particular
significantly weaker for Vic than that for PP at the same initial
protein concentration (Table 5). As reported by O'Kane et al. (2005),
the higher convicilin content in Vic than that in PP was supposed to
Fig. 6. Frequency-dependence at 20  C of storage modulus (G0 ) and loss modulus (G00 )
for pre-heated/aggregated PP, fractions Vic and Leg samples at 4, 4 and 3.5 wt% initial
protein concentration, respectively, at the end of the acid-induced gelation (about 4 h
incubation with GDL, pH z4.5e4.3).
decrease protein aggregation via non-covalent interactions upon
thermal denaturation, and probably during acidification of the
cold-set gelation procedure (Table 3).
4. Conclusion
A GDL-induced cold-set gelation procedure was applied to pea
globulins extracted by ultrafiltration. Protein denaturation
occurred in a range of 75e85  C, while the S
/SeS interchange
reactions occurred between partially-denatured legumin subunits.
The legumin fraction in this study could produce insoluble macro-
aggregates that were detrimental for acid gel formation, whereas
the presence of vicilin and convicilin in the PP sample would alle-
viate the propensity of denatured legumin to form coagulum.
Acid gelation of preheated PP was triggered at pH higher than 6,
even at low protein concentrations; given that predominance of
non-covalent interactions within and between pea proteins ther-
mal aggregates, those would be easily destabilized upon acidifica-
tion, and their local rearrangements into large clusters would led to
a rapid increase in gel strength. Since it occurred in the early stages
of acidification, rapid pea proteins gelation at low temperatures
would provide evidence of the versatility of preheated pea proteins
as an alternative food ingredient.
Acknowledgments
The Plateau de Rheologie des Materiaux Biologiques is thanked
for its technical support. The stakeholders in the food field that are
involved in a proactive approach to change food consumption
models are as well gratefully acknowledged.
References
Alting, A. C., Hamer, R. J., de Kruif, C. G., & Visscher, R. W. (2003). Cold-set globular
protein gels: interaction, structure and rheology as a function of protein con-
centration. Journal of Agricultural and Food Chemistry, 51, 3150e3156.
Boye, J., Zare, F., & Pletch, A. (2010). Pulse proteins: processing, characterization,
functional properties and application in food and feed. Food Research Interna-
tional, 43, 414e431.
Bryant, C. M., & McClements, D. J. (1998). Molecular basis of protein functionality
with special consideration of cold-set gels derived from heat-denatured whey.
Trends in Food Science and Technology, 9, 143e151.
Chavan, U. D., McKenzie, D. B., & Shahidi, F. (2001). Functional properties of protein
isolates from beach pea (Lathyrus maritumus L.). Food Chemistry, 74, 177e187.
Choi, S.-M., & Ma, C.-Y. (2005). Conformational study of globulin from common
buckwheat (Fagopyrum esculentum Moench) by Fourier transform infrared
spectroscopy and differential scanning calorimetry. Journal of Agricultural and
Food Chemistry, 53, 8046e8053.
Choi, S.-M., Mine, Y., & Ma, C.-Y. (2006). Characterization of heat-induced aggregates
of globulin from common buckwheat (Fagopyrum esculentum Moench). Inter-
national Journal of Biological Macromolecules, 39, 201e209.
Ellman, G. L. (1959). Tissue sulfhydryl groups. Archives Biochemistry Biophysics, 82,
70e77.
Laemli, U. K. (1970). Cleavage of structural proteins during the assembly of the head
of bacteriophage T4. Nature, 227, 680e685.
, C., & Gue
Larre guen, J. (1986). Large-scale purification of pea globulins: comparison
between six anion exchangers in medium pressure liquid chromatography.
Journal of Chromatography, 361, 169e178.
Lundqvist, J., Barron, J., Berndes, G., Berntell, A., Falkenmark, M., Karlberg, L., et al.
(2007). Water pressures and increases in food & bioenergy demand implica-
tions of economic growth and options of decoupling. In Scenarios of economic
growth and resource demand, Background to the Swedish Environmental Advisory
Council memorandum (Vol. 1).
Ma, C.-M., & Harwalkar, V. R. (1987). Thermal coagulation of oat globulin. Cereal
Chemistry, 64, 212e218.
Marcone, M. F., Kakuda, Y., & Yada, R. (1998a). Salt-soluble seed globulins of
dicotyledonous and monocotyledonous plants e I: isolation, purification and
characterization. Food Chemistry, 62, 27e47.
Marcone, M. F., Kakuda, Y., & Yada, R. (1998b). Salt-soluble seed globulins of
dicotyledonous and monocotyledonous plants e II: structural characterization.
Food Chemistry, 63, 265e274.
Mession, J.-L., Assifaoui, A., Lafarge, C., Saurel, R., & Cayot, P. (2012). Effect of pea
proteins extraction and vicilin/legumin fractionation on the phase behavior in
admixture with alginate. Food Hydrocolloids, 29, 335e346.
 J.-L. Mession et al. / Food Hydrocolloids 46 (2015) 233e243
Mession, J.-L., Sok, N., Assifaoui, A., & Saurel, R. (2013). Thermal denaturation of Pea
globulins (Pisum sativum L.) e molecular interactions leading to heat-induced
protein aggregates. Journal of Agricultural and Food Chemistry, 61, 1196e1204.
O'Kane, F., Happe, R. P., Vereijken, J. M., Gruppen, H., & van Boekel, M. A. J. S.
(2004a). Characterization of pea vicilin. 1. Denoting convicilin as the a-subunit
of the Pisum Vicilin Family. Journal of Agricultural and Food Chemistry, 52,
3141e3148.
O'Kane, F., Happe, R. P., Vereijken, J. M., Gruppen, H., & van Boekel, M. A. J. S.
(2004b). Characterization of Pea Vicilin. 2. Consequences of compositional
heterogeneity on heat-induced gelation behavior. Journal of Agricultural and
Food Chemistry, 52, 3149e3154.
O'Kane, F., Happe, R. P., Vereijken, J. M., Gruppen, H., & van Boekel, M. A. J. S.
(2004c). Heat-induced gelation of pea legumin: comparison with soybean
glycinin. Journal of Agricultural and Food Chemistry, 52, 5071e5078.
O'Kane, F. E., Vereijken, J. M., Gruppen, H., & van Boekel, M. A. J. S. (2005). Gelation
behavior of protein isolates extracted from 5 cultivars of Pisum sativum L.
Journal of Food Science, 70, 132e137.
Shand, P. J., Ya, H., Pietrasik, Z., & Wanasundara, P. K. J. P. D. (2007). Physicochemical
and textural properties of heat-induced pea protein isolate gels. Food Chemistry,
102, 1119e1130.
243
Shewry, P. R., Napier, J. A., & Tatham, A. S. (1995). Seed storage proteins: structure
and biosynthesis. The Plant Cell, 7, 945e956.
Sun, X. D., & Arntfield, S. D. (2010). Gelation properties of salt-extracted pea protein
induced by heat treatment. Food Research International, 43, 509e515.
Sun, X. D., & Arntfield, S. D. (2011). Gelation properties of salt-extracted pea protein
isolate induced by heat treatment: effect of heating and cooling rate. Food
Chemistry, 124, 1011e1016.
Sun, X. D., & Arntfield, S. D. (2012). Molecular forces involved in heat-induced pea
protein gelation: effects of various reagents on the rheological properties of
salt-extracted pea protein gels. Food Hydrocolloids, 28, 325e332.
Taherian, A. R., Mondor, M., Labranche, J., Drolet, H., Ippersiel, D., & Lamarche, F.
(2011). Comparative study of functional properties of commercial and mem-
brane processed yellow pea isolates. Food Research International, 44,
2505e2514.
Tzitzikas, E. N., Vincken, J.-P., de Groot, J., Gruppen, H., & Visser, R. G. F. (2006).
Genetic variation in pea seed globulin composition. Journal of Agricultural and
Food Chemistry, 54, 425e433.
USDA. (2014). A reference guide to important soybean facts & figures. Soystats.com.
Website.