
compounds from a liquid containing macromolecules (Harrison & Hills,
1997). This model was applied to describe avor retention in whey and
sodium caseinate dispersions (Viry et al., 2018). For esters and alcohols,
good predictions were obtained by only assuming hydrophobic in-
teractions to explain avor retention (Viry et al., 2018). For aldehydes,
stronger retention was observed, which was attributed to the specic
covalent interactions that proteins and aldehydes can undergo (Viry
et al., 2018). Therefore, the model was extended with a covalent inter-
action parameter to describe aldehyde retention (Viry et al., 2018).
Recently, this model was applied to predict avor retention in four plant
proteins and whey dispersions with esters, and ketones, assuming hy-
drophobic interactions only (Snel et al., 2023). Apart from esters and
ketones, aldehydes are an important chemical class to consider for the
avoring of plant-based products such as meat analogues. Therefore,
probing the interactions between plant proteins and aldehydes is
essential and could give great insights into the applicability of the model
when covalent interactions are involved. Furthermore, it would high-
light the relative contribution of hydrophobic and covalent interactions
for aldehyde retention.
This study describes the retention of aldehydes by plant proteins and
applies a avor partitioning model to analyze the results. The proteins
studied are pea protein isolate (PPI), soy protein isolate (SPI), chickpea
protein isolate (CPPI), and fava bean protein isolate (FBPI). Besides,
whey protein isolate (WPI) will be included as a control. The investi-
gated avors include a series (C4, C6, C8, C10) of saturated aldehydes
and 2-mono-unsaturated aldehydes (alkenals), from now on addressed
as aldehydes and alkenals. Furthermore, the obtained partitioning pa-
rameters will be correlated with amino acid composition.
2. Theory: avor partitioning models
Harrison and Hills (1997) developed a rst-order mathematical
model to predict avor release from an aqueous solution containing
polymers at equilibrium conditions. In the case of proteins, avors can
interact with the proteins through either hydrophobic interactions or
specic covalent interactions. The avor-partitioning model will be
shortly summarized here. The partition coefcient (Kf
wg)at equilibrium
between avor concentration in the water phase (ce
fw)and gas phase (ce
fg)
is dened as:
Kf
wg =ce
fg
ce
fw
(1)
When protein is added to the water phase, part of the avors could
interact with the protein. When we consider that the avor-protein
interaction is a reversible, rst-order reaction, the global interaction
constant (Kf
p)between protein P and avor F is dened as:
Kf
p=ce
fp
ce
pce
fw
(2)
in which ce
fp and ce
p are the concentrations of protein-retained avor in
the dispersion at equilibrium, and protein. Since in the experimental set-
up, protein concentration exceeds the avor concentration largely, c
p
is
simplied as the total concentration of protein in the dispersion that
thus remains constant during the experiment. Now, the effective parti-
tion coefcient Keff
wg between avor in the gas phase and the water-
protein phase becomes:
Keff
wg =ce
fg
ce
ft
(3)
in which ce
ft is the total avor in the water system. The mass balance
reads:
cft =cfp +cfw (4)
In which c
ft
equals c
fw
when no protein is present in the water phase.
Using the mass balance and eq. (1) and eq. (2), we can describe Keff
wg as:
Keff
wg =Kf
wg
1+Kf
pce
p
(5)
When avor retention is dominated by hydrophobic interactions, we
could approach the interaction constant with:
Kf
p=apPf
ow (6)
in which a
p
and Pf
ow are the hydrophobic interaction parameter and the
octanol-water partition coefcient. For aldehydes, the covalent inter-
action has to be taken into account, leading to (Viry et al., 2018):
Kf
p=apPf
ow +KAld (7)
in which K
Ald
is the covalent interaction parameter between aldehydes
and proteins. For alkenals, this parameter becomes K
alk
. Aldehydes can
interact with proteins through a condensation reaction (Schiff base
adduct), and alkenals can have an additional conjugate addition
(Michael adduct, Fig. 1) (Anantharamkrishnan et al., 2020a). The total
contribution of both reactions is captured in the parameters K
ald
or K
alk
.
3. Methods and materials
3.1. Materials
Soy protein isolate (SPI, Supro® 500E A) was obtained from Solae
(St. Louis, United States). Pea protein isolate (PPI, Nutralys® F85M) was
obtained from Roquette Fr`
eres S.A. (Lestrem, France). Fava bean protein
isolate (FBPI, FFBP-90-C-EU) and chickpea protein isolate (CPPI, FCPP-
70) were both obtained from AGT Foods (Regina, Canada). Whey pro-
tein isolate (WPI, BiPRO) was obtained from Davisco Foods Interna-
tional (Minnesota, USA). Amino acid content was measured in a
Fig. 1. Schematical representation of the chemical reactions possible between
the amine group of amino acids and butanal (A), and trans-2-butenal (B), and
the thiol group of cysteine with trans-2-butenal (C) (Anantharamkrishnan &
Reineccius, 2020b). Butanal and trans-2-butenal are chosen in this example to
represent aldehydes and alkenals respectively. SB =schiff base, MA =
michael adduct.
S.J.E. Snel et al.