Вы находитесь на странице: 1из 12

Food Hydrocolloids 44 (2015) 208e219

Contents lists available at ScienceDirect

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

Pilot plant preparation of light-coloured protein isolates from de-oiled


sunflower (Helianthus annuus L.) press cake by mild-acidic protein
extraction and polyphenol adsorption
Claudia Pickardt a, b, *, Peter Eisner b, Dietmar R. Kammerer a, c, Reinhold Carle a, d
a
Hohenheim University, Institute of Food Science and Biotechnology, Plant Foodstuff Technology, Garbenstrasse 25, 70599 Stuttgart, Germany
b
Fraunhofer Institute for Process Engineering and Packaging (IVV), Giggenhauser Strasse 35, 85354 Freising, Germany
c
WALA Heilmittel GmbH, Department of Analytical Development & Research, Section Phytochemical Research, Dorfstraße 1, 73087 Bad Boll/Eckwa €lden,
Germany
d
King Abdulaziz University, Biological Science Department, Jeddah 21589, Saudi Arabia

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

Article history: The aim of this study was to produce light-coloured protein isolates from sunflower press cake on a pilot
Received 30 October 2013 plant scale. Mild-acidic extraction at pH 6 prevented discolouration due to co-extracted phenolic com-
Accepted 9 September 2014 pounds and facilitated the adsorptive removal thereof using a styrene-divinylbenzene resin, coupled
Available online 2 October 2014
with ion exchange. To enhance protein solubility under acidic conditions, NaCl was added. Two con-
centrations (2 and 1.3 mol/L) were compared, previously identified as being most suitable for protein
Keywords:
extraction and polyphenol adsorption.
Sunflower protein isolate
Protein isolates recovered by precipitation accumulated globular helianthinin, while highly soluble
Adsorption
Chlorogenic acid
albumins were additionally recovered by ultrafiltration of the supernatants. Precipitated protein yields of
Functional properties 23e26% were obtained when 2 mol/L NaCl solution was used for extraction, whilst the lower salt level
De-oiled sunflower cake only gave protein yields of 15%. Albumin concentrates only marginally added to the overall yield with up
By-product valorisation to 5%.
The physicochemical and functional properties of the precipitated proteins obtained at the different
salt levels were comparable, being slightly inferior for the products obtained on a pilot plant compared to
laboratory scale. Generally, protein isolates obtained by isoelectric precipitation were of high purity and
light in colour, with protein contents of >94% and chlorogenic acid contents of <0.2%. Despite their poor
solubility, they had fair emulsifying and excellent foaming properties.
Altogether, sunflower protein isolates produced according to the novel process are promising food
ingredients. The study demonstrated the feasibility of the process on a pilot plant scale. Moreover, the
simultaneous recovery of phenolic compounds may enhance the economic viability of the overall
process.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction high protein content, the low amounts of antinutritive compounds


and freedom from toxic substances (Gonza lez-Pe
rez & Vereijken,
By-products of plant food processing can be valuable sources of 2007; Kausar, Ali, Javed, & Javed, 2004). The use of sunflower oil
compounds with favourable technological and/or nutritional by-products for the production of food proteins would answer the
properties (Schieber, Stintzing, & Carle, 2001). Press cakes and needs of the food industry which is constantly searching for alter-
meals originating from sunflower oil extraction are promising raw- native food proteins to soy. However, such by-products have until
materials for food proteins due to their widespread availability, the now only been used as low-value animal feed. This has mainly been
due to the poor sensory quality in terms of the dark colour, besides
bitter taste and astringency of sunflower by-products, which can be
* Corresponding author. Fraunhofer Institute for Process Engineering and Pack- ascribed to the presence of phenolic compounds (Gonz alez-Pe rez &
aging (IVV), Giggenhauser Strasse 35, 85354 Freising, Germany. Tel.: þ49 8161 491 Vereijken, 2007). In contrast, the food industry prefers bland and
429; fax: þ49 8161 491 444.
E-mail addresses: claudia.pickardt@ivv.fraunhofer.de, claudia.pickardt@gmx.de
neutral functional ingredients that can be used for the full range of
(C. Pickardt). food applications. However, with regard to their use as food

http://dx.doi.org/10.1016/j.foodhyd.2014.09.020
0268-005X/© 2014 Elsevier Ltd. All rights reserved.
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 209

ingredients, isolated sunflower proteins have been shown to impair the colour and functional properties of sunflower protein
exhibit functional properties partly similar to soy proteins, preparations. Conventional protein extraction under alkaline con-
depending on the isolation procedures used (Brueckner & Mieth, ditions, which is most suitable due to the solubility profile of
1984; Gonza lez-Pe
rez & Vereijken, 2007). Overall, literature data sunflower proteins, results in rapid oxidation of phenolic com-
on the functional properties of sunflower proteins are contradictory pounds and their concomitant covalent bonding to protein side
due to the diversity of methods and pre-treatments employed, as chains (Cater, Gheyasuddin, & Mattil, 1972; Gonza lez-Perez &
well as the large variety of products investigated. Special attention Vereijken, 2007; Pawar, Patil, Sakhale, & Agarkar, 2001). For this
has been paid to emulsifying and foaming properties indicating reason, over the past four decades, processes have been investi-
their potential use as emulsifiers and foaming agents, while their gated to isolate sunflower proteins devoid of phenolic acids.
gelation properties are poor compared to other protein sources. Common strategies include the extraction of sunflower meal with
Sunflower proteins have been shown to form stable emulsions mixtures of organic solvents and/or water to lower the polyphenol
(Gonza lez-Pe rez et al., 2005), while slightly higher emulsifying content prior to alkaline protein extraction, the exclusion of oxygen
capacities were reported for an albumin-enriched isolate compared using vacuum or inert gases, and the use of antioxidants during
to a globulin isolate (Canella, Castriotta, Bernardi, & Boni, 1985). The alkaline protein extraction from polyphenol-rich material and the
emulsifying properties of sunflower proteins have been shown to subsequent removal of co-extracted phenolic acids from the pro-
be slightly inferior, similar and even better than comparable soy tein extracts. These processes have been reviewed by Gonza lez-
protein products in various studies (Brueckner & Mieth, 1984; Perez et al. (Gonza lez-Pe
rez et al., 2002; Gonza lez-Pe
rez &
Gonz alez-Perez & Vereijken, 2007). Shchekoldina and Aider Vereijken, 2007). However, none of these processes has to date
(2012) reported that the emulsifying capacity of a sunflower pro- been transferred to industrial practice, most probably due to the
tein isolate was similar to a soy protein isolate and better than that tedious and inefficient processing. The use of organic solvents may
of wheat gluten, skim milk powder and egg powder. Gonza lez- impair the subsequent protein isolation by diminishing protein
Perez and Vereijken (2007) concluded that “the emulsifying solubility due to denaturation or residual solvent and, in addition,
properties show very interesting perspectives to enhance the usage requires special safety precautions. In contrast, pre-extraction of
of sunflower proteins as they seem to be at least comparable to phenolics with water appeared to be more appropriate for the
those of soy proteins”. Sunflower proteins have also been shown to subsequent protein extraction. However, this measure caused high
form stable foams, although literature reports are also contradic- loss of water-soluble albumins, and proved to be inefficient when
tory and quantitative comparison is virtually impossible due to the followed by alkaline extraction (Salgado, Drago, et al., 2012).
different methods and products. High foaming activities in the Moreover, only few studies have been performed with regard to
range of egg white have been reported (Kabirullah & Wills, 1982) large-scale application of processes. This might be another reason
recommending sunflower proteins as whipping agents. Raymond, why the implementation of such processes has not been pursued.
Rakariyatham, and Azanza (1985) reported the foaming proper- Salgado and co-workers demonstrated the production of sunflower
ties of a sunflower protein preparation to be superior to that of soy protein concentrates with high solubility on a pilot plant scale,
proteins. Lower foaming activities but higher foam stabilities of using sunflower oil cake as the starting material (Salgado, Drago,
sunflower proteins compared to soy protein and overall better et al., 2012) and compared sunflower protein concentrates with
foaming properties compared to skim milk powder and egg powder different contents of phenolic compounds (Salgado, Ortiz,
were reported by Shchekoldina and Aider (2012). Lower foaming Petruccelli, & Mauri (2012) revealing different effects on func-
activities but higher foam stabilities of helianthinin compared to tional properties). However, this study was not focused on col-
sunflower albumins were found by Gonza lez-Perez et al. (2005a). ourless products. Moreover, only a low meal-to-solvent ratio of
Protein solubility has been investigated extensively since it is 1:15 was tested in the previous studies, whereas higher ratios
considered a prerequisite for other functional properties (Saeed & might be favourable with regard to industrial processing (Pickardt,
Cheryan, 1988; Gonza lez-Perez & Vereijken, 2007). Thus, correla- Hager, Eisner, Carle, & Kammerer, 2011). Davin et al., 1983 tested
tions between solubility and emulsifying activity have been re- various patented processes for the preparation of sunflower protein
ported (Karayannidou et al., 2007; Minones Conde, Yust Escobar, isolates on a pilot plant scale and concluded that various qualities
Pedroche Jimenez, Millan Rodriguez, & Rodriguez Patino, 2005) could be produced with higher purities being associated with lower
and increased foaming activities in the presence of salt and after yields. However, functional characterisation of the products has not
hydrolysis were reported (Kabirullah & Wills, 1981; Karayannidou been performed. Taha, El-Nockrashy, & Shoeb (1981) investigated
et al., 2007) due to enhanced solubility. While sunflower albu- the use of counter-current extraction and isoelectric precipitation
mins are known to be readily soluble, most globulin isolates have of sunflower proteins for the preparation of sunflower protein
lower solubilities which are highly pH dependent (Canella et al., isolates in high yield and purity in order to demonstrate suitable
1985; Gonza lez-Pe
rez, Vereijken, Vereijken, van Koningsveld, conditions for industrial operation. However, this study was per-
Gruppen, & Voragen, , 2005b; Kabirullah & Wills, 1982; Minones formed on a laboratory scale, and gently pre-processed sunflower
Conde et al., 2005). However, the values vary significantly and meal was used.
this has been ascribed to differences in the composition and pro- An alternative process for the production of light-coloured
cessing (Brueckner & Mieth, 1984; Gonza lez-Pe
rez et al., 2002; sunflower protein isolates has recently been proposed using
Saeed & Cheryan, 1988) and also different conditions for the solu- mild-acidic protein extraction and high salt concentrations to
bility measurements themselves (Brueckner & Mieth, 1984; enhance the protein solubility to exploitable levels (Pickardt et al.,
Gonz alez-Perez et al., 2002). In particular, thermal or chemical 2009). This process was shown to hinder spontaneous oxidation of
treatment during processing can lead to extensive denaturation co-extracted phenolic compounds, thus facilitating their subse-
(Davin, Berot, & Petit, 1983; Kausar, Ali, Javed, & Khan, 2002). quent removal and recovery by adsorption processes (Weisz,
Furthermore, the binding of phenolic compounds is believed to be Schneider, Schweiggert, Kammerer, & Carle, 2010). The simulta-
responsible for reduced solubility of sunflower proteins (Gonza lez- neous recovery of polyphenols for other uses may improve the
Perez & Vereijken, 2007). economic viability of the overall process (Weisz, Carle, &
Sunflower seeds contain 1e4% phenolic compounds (Gonza lez- Kammerer, 2013). Adsorption conditions have been optimised on
Perez et al., 2002; Kausar et al., 2004; Weisz, Kammerer, & Carle, a laboratory scale to minimise the discolouration and the content of
2009), predominantly chlorogenic acids, which are likely to chlorogenic acids in protein extracts. The combination of both
210 C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219

processes has been successfully performed on a laboratory and


pilot plant scale (Pickardt et al., 2011; Weisz et al., 2010). However,
hitherto only optimal conditions for polyphenol adsorption have
been evaluated on a pilot plant scale (Weisz et al., 2009; Weisz
et al., 2010), while optimisation of the protein recovery has not
been considered thus far. In a previous laboratory study, protein
yields have been shown to increase at higher protein concentra-
tions and elevated salt levels (Pickardt et al., 2011, 2009).
The aims of the present study were to produce light-coloured
protein isolates from sunflower press cake by combining mild-
acidic protein extraction at elevated NaCl concentrations and
adsorptive removal of phenolic compounds using a styrene-
divinylbenzene copolymer resin. The feasibility of the overall pro-
cess should be demonstrated on a pilot plant scale. Furthermore,
two salt levels previously found to be optimal for the individual
process steps of protein extraction and polyphenol adsorption
should be compared in order to test their suitability for the com-
bined process providing evidence of potential process optimisation.
Therefore, the processes and products should be compared on a
laboratory and pilot plant scale in order to identify possible effects
of the different processing conditions. Relevant functional proper-
ties of the derived protein isolates were determined in order to
indicate their suitability as food ingredients. As the isoelectric
precipitate has been shown to solely consist of the globulin fraction
(Pickardt et al., 2011), the recovery of the water-soluble albumins
from the supernatants was included in the pilot plant experiments
in order to demonstrate ways allowing total protein recovery.
Fig. 1. Preparation of sunflower protein isolates (IP) and concentrates (UF) on a lab-
2. Materials and methods oratory (L) and pilot plant (P) scale. Individual parameters are listed in Table 1.

2.1. Materials

Cold-pressed sunflower cakes from industrial sunflower oil respectively, as indicated in Table 1, and at a solid-to-liquid ratio of

processing were provided by Teutoburger Olmühle (Ibbenbüren, 1: 10 (1 kg de-oiled cake þ 9 kg NaCl solution). The exceptions were
Germany). These cakes were de-oiled with iso-hexane in a pilot IP-P1 and UF-P1, where a ratio of 1:15 was applied. These condi-
plant percolator (volume 1500 L, e&e Verfahrenstechnik, Ware- tions were chosen based on preliminary studies on protein
ndorf, Germany) and flash-desolventised with hexane vapour at extraction (Pickardt et al., 2011) and polyphenol adsorption (Weisz
400e500 mbar, prior to steam desolventisation at a maximum et al., 2010). Laboratory extraction was performed using a volume of
product temperature of 60  C. Prior to protein extraction on a 10e12 L, while pilot plant extraction was scaled up to 1500 (P1) and
laboratory scale, the de-oiled cake was milled in a pin mill (Alpine, 1800 L (P2) using the equipment suggested by D'Agostina et al.
Augsburg, Germany) and passed through a sieve of 0.5 mm mesh (2006). After extraction for 60 min at room temperature
size. The de-oiled cake contained 38.1 ± 2.7 g protein per 100 g (20 ± 3  C) and constant pH of 6.0 by adjusting with HCl or NaOH
(N  5.6), 7.1 ± 0.8 g ash per 100 g, 2.1 ± 0.5 g fat per 100 g, (1 M on a laboratory scale and 3 M on a pilot plant scale), as
2.1 ± 0.7 g chlorogenic acid (CGA) per 100 g, and 51 ± 2 g dietary necessary, under gentle agitation (150 rpm on a laboratory scale
fibres and sugars per 100 g, relative to dry matter (dm, 90.3 ± 0.7 g and 50 rpm on a pilot plant scale, the crude protein extracts were
per 100 g). The methods used are listed below (Section 2.3.1). recovered by centrifugation (a laboratory beaker centrifuge at
Food grade NaCl was purchased from a local supermarket. 3300 g and a continuously running decanter (~1500 L/h) at approx.
Methanol (Chromanorm for HPLC) was from VWR International 3000 g on a pilot plant scale)). Phenolic compounds were removed
(Darmstadt, Germany), chlorogenic acid (98%) was obtained from from the crude extracts by adsorption onto a styrene-
Sigma-Aldrich (Steinheim, Germany), ortho-phosphoric acid (85%) divinylbenzene resin Amberlite XAD 16HP (Rohm & Haas,
from Merck (Darmstadt, Germany), and HPLC gradient grade water Chauny, France) as specified in Table 1. For the preparation of iso-
from J.T. Baker (Deventer, Netherlands). All other reagents used lates IP-L1 and IP-L2, 400 and 500 g of preconditioned adsorbent
were of analytical grade. Reference products Supro 500E and Supro resin, respectively, were added to 7 and 9 L of the crude protein
EX33 were obtained from Protein Technologies International (now: extract and gently agitated (~150 rpm) for 4 h at 20  C under a
Danisco Dupont), and ProFam 464 from ADM Specialty Ingredients, nitrogen blanket in an agitator vessel as previously described
Koog aan de Zaan, The Netherlands. (Pickardt et al., 2011). Subsequently, the resin was removed using a
Miracloth filter (rayon-polyester, Merck, Darmstadt, Germany). For
2.2. Isolation of sunflower proteins by mild-acidic protein the preparation of isolates IP-P1 and IP-P2 and concentrates UF-P1
extraction and polyphenol adsorption followed by isoelectric and UF-P2 on a pilot plant scale (see Table 1), a two-step continuous
precipitation or ultrafiltration procedure was applied combining ion exchange and adsorption:
The protein extract was pumped through two cylindrical steel
Protein isolation was performed according to the protocol vessels connected in series, packed with 90 L of an anion exchange
depicted in Fig. 1 with the parameters as detailed in Table 1. For resin (Lewatit S 6328 A; Lanxess, Leverkusen, Germany) and the
protein extraction, the de-oiled cake was immersed in salt solutions aforementioned adsorbent resin, respectively, at a flow rate of 3 bed
having NaCl concentrations (cNaCl) of 1.3 and 2.0/2.1 mol/L, volumes per hour as described by Weisz et al. (2010). Resin
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 211

Table 1
Sample assignment and processing parameters for protein isolation.

Sample Characteristic processing steps Scale cNaCl (mol/L) Polyphenol removal pH of Ultra- Washing procedure
code during protein precipitation filtration
extraction

IP-L1 Saline extraction, polyphenol adsorption, Lab 1.3 Adsorber resin 3.5  1) 1/5, pH 4.0, 10  C
isoelectric precipitation 2) 1/5, pH 6.0, 10  C
IP-P1 Saline extraction, polyphenol adsorption, Pilot 1.3 Ion exchange þ adsorber resin 3.2  1) 1/4, pH 3.9, 15  C
isoelectric precipitation 2) 1/5, pH 4.1, 15  C
UF-P1 Saline extraction, polyphenol adsorption, Pilot 1.3 Ion exchange þ adsorber resin (3.2) þ Diafiltration vs. tap water
ultrafiltration of supernatant after
precipitation

IP-L2 Saline extraction polyphenol adsorption, Lab 2.0 Adsorber resin 3.5  1) 1/4, pH 4.0, 10  C
isoelectric precipitation 2) 1/4, pH 6.0, 10  C
IP-P2 Saline extraction, polyphenol adsorption, Pilot 2.1 Ion exchange þ adsorber resin 3.3  1) 1/4, pH 3.5, 15  C
isoelectric precipitation 2) 1/5, pH 4.5, 15  C
UF-P2 Saline extraction, polyphenol adsorption, Pilot 2.1 Ion exchange þ adsorber resin (3.3) þ Diafiltration vs. tap water
ultrafiltration of supernatant after
precipitation

activation was performed using ethanol (60%, v/v) followed by system (TGA 601, Leco Instrumente, Mo €nchengladbach, Germany)
NaOH (2.5 mol/L) and deionised water (50  C) as described previ- using standard methods (AOAC, 2005a). The protein content was
ously (Weisz et al., 2009). Prior to use, the resin was preconditioned calculated from the nitrogen content as determined by the Dumas
with NaCl solutions of the respective concentration. combustion method (AOAC, 2005b), using a Protein/Nitrogen
To recover protein isolates (IP), the proteins were precipitated Analyzer FP 528 (Leco, St. Joseph, MI, USA) and applying a con-
by adding HCl (1e3 mol/L, depending on the required amounts) to version factor of 5.6, which has been found to be most suitable for
reach the pH values listed in Table 1. The protein curd obtained was sunflower proteins based on their amino acid profile as determined
separated by centrifugation in a beaker centrifuge on a laboratory in our laboratory (data not shown). Similar factors of 5.7 and 5.5, for
scale and in a continuous plate separator (~500 L/h) on a pilot plant example, were used by Canella et al. (1985) and Zilic et al. (2010)
scale and washed twice by mixing with water at ~ 15  C at a mass respectively. The lipid content was quantified by gas chromatog-
ratio of 4e5: 1 as detailed in Table 1. The washed curd was sepa- raphy after extraction with butanol and saponification (DGF, 2011).
rated as described above, neutralised to pH 6.7e7.0 by adding NaOH The carbohydrate content (dietary fibres and sugars) was deter-
solution, homogenised using a colloid mill and spray dried in a Niro mined by the difference of total dry matter contents and those of all
atomizer (2 L/h) on a laboratory scale and in a GEA atomizer (25 L/ other compounds.
h) on a pilot plant scale, both from GEA (Mühlheim, Germany). Both
dryers were equipped with a mixing nozzle of 0.7 mm inner 2.3.2. Quantification of phenolic acids by HPLC-DAD
diameter. They were operated at an inlet temperature of Phenolic acids were determined by HPLC with UV detection at
170e180  C and outlet temperature of 75 ± 3  C. On a pilot plant 330 nm based on the method of Thiyam, Sto € ckmann, Zum Felde,
scale, the feed pipe was equipped with a heat exchanger, effecting a and Schwarz (2006). Only chlorogenic acid (CGA; 5-caffeoylquinic
short-duration pasteurisation at 55 ± 2  C. acid) was quantified as lead substance due to its predominance in
To recover soluble proteins (UF), the supernatants after precip- sunflower seeds (Weisz et al., 2009). Aliquots of 1 g of the dry
itate separation were subjected to ultrafiltration using a membrane samples were extracted in triplicate with 8 mL of 0.29 mol/L HCl,
of 10 kDa cut-off as described by D'Agostina et al. (2006) at a sonicated for 1 min and centrifuged (20  C, 10 min, 20,000  g).
transmembrane pressure of 0.2 MPa and a volume concentration Extracts were filtered through folded Whatman filters and pooled
ratio of 20. Co-solutes were separated by semi-continuous diafil- in volumetric flasks. The total volume was made up to 25 mL.
tration using a diafiltration volume of 5, and the concentrated 600 mL of the extract were mixed with 1400 mL of methanol in
protein solution was pasteurised inside the ultrafiltration vessel 2.0 mL Eppendorf tubes and centrifuged for 5 min at 12,100  g to
before spray drying as described above. remove precipitated proteins. The samples were appropriately
Product and protein yields were calculated based on the dry diluted with 70% methanol (v/v), and filtered using syringe filters
matter and protein contents of the isolates prior to drying. This (0.45 mm). HPLC analyses were carried out using an HPLC pump
ruled out variable and disproportionally high losses associated with P680, an automated sample injector ASI-100, a column oven ICS-
batch spray drying, since optimization of the spray drying step was 3000 DC, and a UV/VIS detector UVD170U operated by Chrome-
not part of our study. Nevertheless, comparison with other studies leon software (Dionex Softron, Germering, Germany). The column
seems appropriate as most studies use freeze drying or air drying in used was a Synergi Fusion RP (250  4.6 mm) equipped with a
cups which do not cause significant losses either. While the product guard column cartridge with C18 polar embedded material of 5 mm
yield represents the percentage of dry matter recovered as protein particle size (Phenomenex, Aschaffenburg, Germany), operated at
isolate relative to the initial dry matter content of the de-oiled cake, ambient temperature. Gradient elution was performed using water/
the protein yield represents the amount of protein in the protein methanol (90/10, v/v) with 0.2% ortho-phosphoric acid as eluent A
isolate relative to the protein content of the de-oiled cake. and methanol with 0.1% ortho-phosphoric acid as eluent B. The
linear gradient programme was as follows: 10% B (0 min), 20% B
2.3. Analytical methods (7 min), 45% B (20 min), 70% B (25 min) and 100% B (28 min). The
injection volume was 20 mL, and all determinations were performed
2.3.1. Proximate analysis of de-oiled press cake and protein isolates in duplicate. For calibration, a chlorogenic acid stock solution of
The dry matter and ash contents were determined gravimetri- 2 mg/mL in 70% methanol was diluted with the same solvent to
cally at 105  C and 550  C, respectively, in a thermogravimetric obtain concentrations in the range of 0.02e0.5 mg/mL.
212 C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219

2.4. Gel electrophoresis (SDS capillary PAGE) parameters were not optimised for the specific process. However,
the aim of the pilot plant trials was to demonstrate the feasibility of
The protein molecular weight distribution was determined by the process on a pilot plant scale and to recover protein isolates for
sodium dodecyl sulphate capillary polyacrylamide gel electropho- characterisation of their properties. The goal was not to determine
resis (SDS capillary PAGE) as previously described (Pickardt et al., optimum process parameters for system configuration and plant
2011), using a capillary electrophoresis system HPCE 3D (Agilent, layout, nor to carry out precise quantification for an economic
Waldbronn, Germany) equipped with 3D-CE ChemStation software assessment. Thus, yields are given only as an indicator to allow
(Rev. A.08.01) applying the CE-SDS protein kit 148-4160 (Bio-Rad general comparison with the laboratory runs performed at the
Laboratories, Munich, Germany). The BioRad protein size standard same salt concentration.
148-2015, containing molecular weight markers ranging from 14 to
205 kDa (lysozyme (14,400 Da), trypsin inhibitor (21,500 Da), car- 3. Results and discussion
bonic anhydrase (31,000 Da), ovalbumin (45,000 Da), BSA
(66,200 Da), phosphorylase b (97,000 Da), b-galactosidase Dark discolouration due to the oxidation and protein binding of
(116,000 Da), and myosin (200,000 Da)), was used for calibration phenolic acids is a major drawback of sunflower proteins. As light
with benzoic acid as internal migration time reference, which was colour is a prerequisite for versatile industrial application of protein
detected after ~4.5 min. Dithiothreitol was added for reducing isolates, the removal of chlorogenic acid derivatives was an
disulphide bridges. important aspect of the present study. Sunflower protein isolates
were produced from de-oiled press cake to demonstrate the
2.5. Physicochemical characterisation of the protein isolates feasibility of the novel process that combines mild-acidic protein
extraction at pH 6 with adsorptive removal of phenolic compounds
2.5.1. Colour from the protein crude extracts. The process scheme is depicted in
The colour of the dried powdered products was determined Fig. 1. Polyphenol removal was accomplished by a single-step
using the CIE-L*a*b* colour system with a CR-300 Chroma Meter adsorption onto Amberlite resin on a laboratory scale and by
(Minolta, Osaka, Japan) as the mean values of 8 measurements. resin adsorption coupled with ion exchange on a pilot plant scale.
Furthermore, two different NaCl levels (1.3 mol/L and 2 mol/L,
2.5.2. Protein solubility, emulsifying and foaming properties respectively), which were previously found to be suitable for the
Protein solubility at pH 7 was determined according to the individual processing steps of protein extraction and polyphenol
method of Morr et al. (1985) as the proportion of dissolved protein. adsorption, were used in order to compare their suitability for the
Protein solutions were prepared by dissolving 1.5e2.5 g sample in combined process (see Table 1): While the higher level was shown
50 mL of 0.1 mol/L NaCl adjusting the pH for 60 min under constant to best favour protein extraction (Pickardt et al., 2009) and isolation
magnetic stirring (~200 rpm) at ambient temperature and with (Pickardt et al., 2011), the lower level had been demonstrated to be
successive removal of undissolved matter by centrifugation more favourable for the adsorption process onto Amberlite resin
(20,000  g, 15 min). The content of dissolved proteins was ana- (Weisz et al., 2010).
lysed by the Dumas method (see Section 2.3.1).
The emulsifying capacity of the protein isolates was determined 3.1. Composition and purity of the products
as described by Yoshie-Stark, Wada, and Waesche (2008) by
detecting the sudden decrease in conductivity at the point of phase 3.1.1. Protein and ash contents
inversion of the emulsion upon constant addition of oil (corn oil, The major protein fractions were recovered from the extracts by
Mazola®, Unilever Deutschland, Hamburg, Germany) to a 1% isoelectric precipitation (IP) followed by centrifugation. The
dispersion of the protein isolates or concentrates while dispersing resulting supernatant was subjected to ultrafiltration on the pilot
with an UltraTurrax at 24,000 rpm. plant scale, thereby yielding an additional product (UF, see Fig. 1).
The foaming properties of the protein isolates and concentrates The chemical compositions of the different samples are shown in
were determined as described by D'Agostina et al. (2006) using Table 2. Major differences in the protein contents of IP and UF were
200 mL of 5% suspensions of the products in deionised water at pH found. In general, isoelectric precipitation yielded protein isolates
6.7e7.0 after whipping in a Hobart 50 N mixer for 8 min at high of high purity with protein contents ranging from 94.3 to 98.6% of
speed (level 3) at room temperature. The foaming activity repre- the dry matter, with slightly higher protein contents after extrac-
sents the foam overrun and is expressed as the percentage of initial tion at the higher salt level as well as higher yields on a laboratory
volume (D'Agostina et al., 2006). The foam drainage stability was scale compared to the pilot plant scale. Differences between the IP
determined in a 250 mL measuring cylinder as the percentage of isolates were generally insignificant except between IP-P1 and IP-
foam volume remaining after 60 min. L2. Despite the high salt concentrations used during extraction,
the ash contents were low and ranged between 1.3% and 2.0%.
2.6. Data evaluation Significant differences were found between IP-L1 and IP-P1 which
could be ascribed to more efficient washing on a pilot plant scale
All analyses were performed in duplicate, and values are given and between IP-P2 and IP-P1 probably due to higher initial salt
as mean ± mean deviation unless otherwise stated. Statistical dif- content during extraction.
ferences between mean values were analysed by pairwise t-test In contrast, ultrafiltration yielded protein contents ranging from
(P ¼ 0.05) after running an F-test to check for equality of variances. 65.8 to 68.7% (Table 2). Although the respective products are
The protein isolates and concentrates were produced without considered to be protein isolates according to the underlying
repetition on a pilot plant scale. Therefore, product characterisation isolation process, they did not fulfil the requirements of protein
was based on one production batch, the big volumes ensuring isolates, because of their protein contents of <90%. Consequently,
sufficient product homogeneity. The pilot plant was used in a they should be named protein concentrates. The divergent purities
standard configuration that had been proved to provide adequate were also reflected in their increased ash contents ranging from
reproducibility in several previous studies (e.g. Yoshie-Stark et al., 3.3% to 21.6%, which was extremely high.
2008; D'Agostina et al., 2006). Due to the high cost it was not The protein contents of the precipitated proteins were in the
possible to repeat pilot plant runs in the present study, and the same range as reported for other sunflower protein and
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 213

Table 2
Purity and colour of sunflower protein isolates (IP) and concentrates (UF) prepared on a laboratory (L) and pilot plant (P) scale.

Sample codea Dry matter (dm) % Protein (N  5.6) % dm Ash % dm CGA % dm Colour parameters

L* a* b*
b
IP-L1 95.0 ± 1.0d 97.1 ± 0.5d,e 2.0 ± 0.1d n.d. 84.7 ± 0.8d 1.6 ± 0.0d,h 12.5 ± 0.4d,g
IP-P1b 94.7 ± 0.9d 94.3 ± 0.5d 1.3 ± 0.0e 0.02 ± 0.01d,e 82.5 ± 0.8d,e 1.7 ± 0.1d,e 11.7 ± 0.4d,e
UF-P1b 92.3 ± 0.9d 68.7 ± 0.3f 3.3 ± 0.1f n.d. 78.7 ± 0.8e,f 2.0 ± 0.1e,g 9.7 ± 0.3f

IP-L2c 92.7 ± 0.9d 98.6 ± 0.5e,g 1.3 ± 0.2d,e,g 0.15 ± 0.03d 86.8 ± 0.9d 0.9 ± 0.0f 10.2 ± 0.3e,f
IP-P2b 94.4 ± 0.9d 96.3 ± 0.5d,g 1.6 ± 0.0g <0.01e 75.1 ± 0.8f 2.4 ± 0.1g 14.2 ± 0.4g
UF-P2b 91.0 ± 0.9d 65.8 ± 0.3h 21.6 ± 0.6h <0.01e 61.6 ± 0.6g 1.4 ± 0.0h 10.2 ± 0.3e,f

CGA e chlorogenic acid, taken as lead substance for phenolic compounds.


deh: Values with the same online letters within one column are not significantly different (p ¼ 0.05).
a
For details refer to Table 1 and Fig. 1. dm e dry matter, n.d. not determined.
b
Mean of at least two determinations of one isolate ± mean deviation.
c
Average of four values (in each case two determinations on two different isolates) ± standard deviation, data from Pickardt et al. (2011).

helianthinin isolates (Canella et al., 1985; Gonza lez-Pe


rez et al., in the pilot plant experiments was below the threshold level of
2002; Kabirullah & Wills, 1981). Similar protein contents were 0.05 g/100 g reported for the perception of discolouration by Saeed
also found in low-polyphenol protein isolates by Kabirullah and and Cheryan (1988), while this limit was exceeded in IP-L2 which
Wills (1982), Pawar et al. (2001), Saeed and Cheryan (1988) and contained 0.15 g CGA/100 g. However, all protein isolates recovered
Shchekoldina and Aider (2012). However, in their studies the ni- by precipitation had very light colours (off-white to cream-
trogen conversion factors used were higher than in our work. The coloured, data not shown). These had high lightness values L* be-
ash contents of the sunflower proteins were in the same range or tween 82.5 and 86.8 except for IP-P2 (L* ¼ 75.1), and low a* and b*
even lower than in the aforementioned studies. values, as specified in Table 2. a* values corresponding to a reddish
The lower purities of the sunflower protein concentrates ob- hue were very low ranging between 0.9 and 2.4 in all products, and
tained by ultrafiltration may be due to insufficient dilution of the b* values corresponding to a faint yellow hue varied between 9.7
initial salt concentration during diafiltration, which is reflected in and 14.2. The UF concentrates had slightly darker (light grey to
the increased mineral contents of UF-P2 recovered from the higher beige) colours with L* values of 78.7 (UF-P1) and 61.6 (UF-P2) (as
salt level compared to UF-P1 obtained from the lower salt level shown in Table 2).
extracts. These findings were accompanied by slight sweet and In general, the novel process efficiently removed CGA and pro-
bitter off-tastes for these products in sensory evaluation (data not duced light-coloured protein isolates. It was as effective in reducing
shown), indicating incomplete removal of co-extracted solutes. The CGA contents and enhancing product colour as was exhaustive
lower purities of protein isolates recovered by ultrafiltration polyphenol extraction with 80% methanol (Gonza lez-Perez et al.,
compared to precipitated proteins were also found by D'Agostina 2002), extraction with acidic butanol prior to alkaline protein
et al. (2006), whereas Yoshie-Stark et al. (2008) and Xu and extraction and precipitation (Pawar et al., 2001; Saeed & Cheryan,
Diosady (2002) reported opposite findings. Phytic acid interfer- 1988), and protein precipitation with succinic acid (Shchekoldina
ence with proteins in the supernatant may be another reason for & Aider, 2012). Similarly, the use of absorber technology in the
the increased ash contents, since this compound was shown to be processing of rape seed proteins resulted in light-coloured products
extracted from sunflower seeds without being completely precip- having low phenolic acid contents and better sensory acceptance
itated (Pickardt et al., 2011). The two salt concentrations in both than products having higher phenolic contents (Xu & Diosady,
supernatants might have favoured phytate complexation during 2002).
ultrafiltration to different extents due to altered charge effects While the isolate IP-L2 produced at an increased salt level had a
(Saeed & Cheryan, 1988). In fact, ultrafiltration is a complex process lighter colour than IP-L1, the product IP-P2 obtained on a pilot plant
requiring further investigation in order to improve the yield and scale was significantly darker than IP-P1 (Table 2), although it had a
purity of soluble proteins recovered from the supernatants. lower CGA content. However, both products recovered on a pilot
plant scale at the higher NaCl concentration (IP-P2 and UF-P2) were
3.1.2. Residual phenolic acid contents and colour of the protein significantly darker than the other products. Apparently, in the case
isolates of P2, the polyphenol adsorption capacity was insufficient. This
Chlorogenic acid (CGA) was determined as the lead substance appears to be in agreement with previous findings indicating that
representing the majority of phenolic compounds in sunflower the lower salt level was most suitable for polyphenol adsorption
seeds (Weisz et al., 2009). After the two-step adsorption, the re- due to the lower extract concentrations (Weisz et al., 2010),
sidual CGA content was below 0.02 g/100 g dry matter (dm) for the although this was not confirmed in laboratory experiments (IP-L2).
protein isolates obtained in the pilot plant experiment (Table 2). In fact, in addition to the higher extract concentration, the amount
This corresponds to CGA removal of 99% based on an initial CGA of extract used in P2 was greater than in P1 (1800 vs. 1500 L) while
content of 2.1 g/100 g of the raw material. In contrast, the residual applying the same amount of adsorbent resin. Hence, more detailed
CGA content after single-step adsorption on a laboratory scale was adaptation would be needed to achieve the same efficiency as on a
0.15% in IP-L2 (Pickardt et al., 2011), corresponding to a reduction of laboratory scale. Moreover, the reduced adsorption efficiency in the
~90%. case of P2 was also aggravated by repeated use of the adsorbent
The CGA content was lower in the products obtained on a pilot resin. To improve polyphenol adsorption, the amount of resin and
plant scale than on the laboratory scale. This was ascribed to contact time may be increased, which has been shown to be as
enhanced removal of monomeric phenolic acids by the ion ex- effective as lowering the load of the protein extract on a laboratory
change resin, whereas the adsorbent resin only lowered the con- scale (Weisz et al., 2010). As another option, the use of an inter-
tents of polymeric phenolic compounds (Weisz et al., 2010). The mediate salt level appears to be promising for optimising the pro-
residual CGA content of 0.02 g/100 g in the protein isolates obtained tein yields while maintaining high adsorption capacity. For
214 C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219

example, 1.8 mol/L NaCl was reported to be most suitable for pro- 41e42 kDa, as well as minor fractions of ~15, 31, 35 and 50 kDa
tein extraction by Shchekoldina and Aider (2012). under reducing conditions (Fig. 2). The electropherograms of pro-
Comparing the precipitated isolates from different scales tein isolates IP-L1 and IP-L2 obtained on a laboratory scale showed
(Table 2), the product colours did not strictly correlate with the identical patterns and were therefore omitted. The fractions with
phenolic acid contents. This agrees with the observations of molecular weights between 22 and 42 kDa represent basic and
Salgado, Drago, et al. (2012), and might be ascribed to the fact that acidic polypeptides derived from the helianthinin subunits by
darkening already occurred during extraction and the subsequent reducing disulphide bonds (Kortt & Caldwell, 1990; Raymond,
separation steps prior to phenolic acid removal. On a pilot plant Inquello, & Azanza, 1991; Zilic et al., 2010). The intact subunits
scale, the longer processing times of higher volumes allowed were detected under non-reducing conditions as the predominant
enhanced polyphenol oxidation and this impaired the colour of the peaks corresponding to molecular weights of 56e58, 66e68, and
final products. Moreover, contact with oxygen was not prevented 70e71 kDa with minor differences between the IP isolates (data not
on a pilot plant scale, whereas laboratory extraction was performed shown), which is in accordance with literature data (Gonza lez-
in an agitator vessel under a nitrogen blanket, and centrifugation rez et al., 2002; Gonz
Pe rez et al., 2002; Kortt & Caldwell,
alez-Pe
was carried out in sealed beakers. The two-step adsorption was 1990; Raymond et al., 1991). In contrast, the electropherograms of
designed to remove monomeric CGA and oxidised derivatives, thus the concentrates UF-P1 and UF-P2 obtained from the supernatants
improving the visual appearance of the proteins, provided that represent the sunflower albumins characterised by fractions of
sufficient amounts of resin are used. 10e17 kDa with the main peak at ~12 kDa. Our findings are in
agreement with literature data (Gonz rez et al., 2005b;
alez-Pe
3.2. Protein characterisation Pandya et al., 2000; Zilic et al., 2010). As previously reported
(Pickardt et al., 2011), an additional fraction was identified at
The molecular weight distributions of the different protein ~23 kDa solely under reducing conditions, thus presumably rep-
preparations are illustrated in Fig. 2. The results for the precipitated resenting residual helianthinin subunits. Moreover, UF-P1 con-
proteins IP-P1 and IP-P2 were very similar, revealing the predom- tained considerable amounts of globulins represented by the peaks
inance of molecular weight fractions of around 22e25, 32e33 and of the subunits at 32e33 and 41e42 kDa detectable under reducing

Fig. 2. Molecular weight distribution of sunflower protein isolates (IP) and concentrates (UF) prepared on a pilot plant scale (P). SDS capillary gel electrophoresis under reducing
conditions; std. ¼ molecular weight standard; for other abbreviations see Table 1 and Fig. 1.
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 215

conditions (Fig. 2, UF-P1), while UF-P2 only contained trace


amounts of these globulins. These results confirm the enrichment
of the globulin fraction in the precipitated isolates and the recovery
of albumins by ultrafiltration, with the higher salt level promoting
precipitation in agreement with our previous findings (Pickardt
et al., 2011).

3.3. Protein yields

The protein yields obtained by precipitation (IP) highly depen-


ded highly on the salt levels, ranging from 14.7 to 15.0% at the lower
salt level to 22.7 and 25.6% at the higher NaCl level on a laboratory
and pilot plant scale, respectively (Fig. 3). This corresponds to
product yields ranging from 6.2 to 10.6%. The protein distribution
throughout the process is illustrated in Fig. 4 taking the IP/UF-P2
run as an example: The protein yields and losses in all product
and by-product streams, respectively, are indicated following the
same trend for all experiments. Fig. 4. Protein distribution in products and waste streams during pilot plant pro-
cessing at cNaCl ¼ 2 mol/L (P2). Protein yields (dark bars) and losses (light bars) of
The protein yields on a pilot plant scale are only indicative due
individual processing steps as shown in Fig. 1 relative to initial sunflower cake protein.
to the single experiments. However, they are in the same range as adsorption incl. ion exchange, ultrafiltration incl. diafiltration.
the laboratory scale yields analogously showing the influence of the
salt concentration, thus confirming the feasibility of the process. As step for polyphenol removal. Similar or slightly higher yields of
previously shown, higher salt levels resulted in higher protein ~30% were reported for phytate and/or polyphenol depleted pro-
concentrations in the extracts (Pickardt et al., 2009), and thus, tein isolates obtained by alkaline extraction and isoelectric pre-
higher efficiency of the precipitation step (Pickardt et al., 2011). In cipitation after pre-extraction with acidic butanol (Saeed &
contrast, the protein yields of the concentrates obtained by mem- Cheryan, 1988) and water (Salgado, Molina Ortiz, Petruccelli, &
brane filtration (UF) were only 1.2% and 5.3%, corresponding to Mauri, 2011) or by extraction with salt solution after acetone
product yields of 0.7% and 3.2% for UF-P1 and UF-P2 respectively. As treatment (Prasad, 1990), while yields of 40% were reported after
precipitation sufficed to recover the major protein fractions, only alkaline extraction and acidic precipitation (Kabirullah & Wills,
small yields were achieved using the subsequent membrane 1981). Alkaline extraction on a pilot plant scale (Davin et al.,
filtration (see Fig. 4). 1983) gave protein yields ranging between 23 and 52%, with an
The protein yields in the precipitated isolates of between 15.0 inverse correlation of yields and protein contents of the isolates. ln
and 25.6% (Fig. 3) and the cumulative total protein yields (PI and contrast, Shchekoldina and Aider (2012) reported high protein
UF) of between 16.2 and 30.9% were similar to the yields reported yields of up to 50% using alkaline extraction combined with 10%
by Salgado, Drago, et al. (2012) of 12e14% and 26e29% with and NaCl (corresponding to 1.8 mol/L) from de-oiled sunflower cake
without precipitation respectively, after alkaline extraction from followed by acidic precipitation, whilst 60% of meal protein was
de-oiled sunflower cake on a pilot plant scale with a pre-extraction recovered by Gonza lez-Perez et al. (2002) in a total sunflower

Fig. 3. Comparison of sunflower protein isolates (IP) and concentrates (UF) after mild-acidic extraction at NaCl levels of cNaCl ¼ 1.3 (1) or 2.0/2.1 mol/L (2) prepared on a laboratory
(L) or pilot plant (P) scale. For sample details refer to Table 1 and Fig. 1. Values of functional properties are given as mean ± mean deviation. Protein yields are indicative and the
deviations of individual processing steps combined. Bars with the same superscript letter within one chart are not significantly different (p ¼ 0.05). *Foams of UF-P1 were not stable
after 60 min.
216 C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219

protein isolate after repeated alkaline extraction and ultrafiltration. however significantly lower solubility. In contrast, albumins
The lower yields in the present study compared to the latter studies recovered by membrane filtration exhibited far better water solu-
may partly be due to the mild-acidic vs. alkaline extraction which bility (72.4 and 74.4% for UF-P1 and UF-P2, respectively).
might result in lower extraction yields despite the high salt levels General differences in the functional properties of the IP protein
used (Pickardt et al., 2009). In fact, the extraction yield was only isolates and UF concentrates may be ascribed to the predominance
about 57% for IP-L2 (Pickardt et al., 2011) and similar at pilot plant of globulins and albumins in the protein preparations obtained by
scale (60% for IP-P2, see Fig. 4). To enhance extraction yields, a precipitation and ultrafiltration respectively (see Section 3.2 and
second extraction step including counter-current extraction ap- Fig. 2). Sunflower albumins are known to be readily soluble,
pears helpful. In contrast, the use of elevated temperatures to in- whereas helianthinin solubility is generally low and pH dependent
crease protein extraction (Pickardt et al., 2009; Shchekoldina & (Canella et al., 1985; Gonza lez-Pe rez et al., 2005b; Kabirullah &
Aider, 2012) is not recommended due to enhanced polyphenol Wills, 1982). Therefore, the small but significant difference be-
binding and decreased adsorption efficiency (Pickardt et al., 2009; tween IP-L1 and IP-P2 might be ascribed to the slightly higher ash
Weisz et al., 2010). Moreover, protein yields also depend on the content (see Table 2), which is likely to interfere in the solubility
quality of the starting material. In contrast to gently processed measurements (Pickardt et al., 2011).
experimental sunflower meal used by Gonza lez-Pe
rez et al. (2002) However, our study revealed extremely poor solubility for the
and Kabirullah and Wills (1981), industrial press cakes were used in protein isolates in water at pH 7, irrespective of the production scale
the present study. Harsh treatments during industrial de-oiling and NaCl level used. This was in agreement with earlier findings
may induce partial denaturation of the proteins (Salgado et al., showing solubility to be affected by the extraction and precipitation
2011) affecting protein solubility (Davin et al., 1983; Kausar et al., conditions which was partly ascribed to denaturation (Pickardt
2002; Raymond et al., 1985; Shchekoldina & Aider, 2012). Even et al., 2011). Protein denaturation has been determined for IP-L2
though low temperatures were used during solvent extraction, the at a denaturation enthalpy of 0.1 J/g protein (main peak at 99  C)
starting material used in the present study may have been affected as compared to 3.1 J/g protein and 2.4 J/g protein for the respective
by such treatment. raw material and the protein isolate obtained by precipitation at pH
Marked protein losses associated with further purification steps 4.5, respectively (Pickardt et al., 2011). This confirms the denaturing
used in the present study (see Fig. 4), are another reason for the effect of low pH values. Similarly, denaturation enthalpy for IP-P2
lower yields compared to the process proposed by Gonza lez-Perez was only 9% of the value measured for the press cake (data not
et al. (2002). Approximately 1/4 of the extracted protein was lost shown). Probably, denaturation of all IP products was high. Most
upon precipitation (Pickardt et al., 2011) which could not be likely, harsher process conditions (longer times, stronger stirring
completely recovered by ultrafiltration. High losses of low molec- and mechanical treatment and local pH dips and slightly lower pH
ular weight proteins in the permeate might be mitigated by using values) caused severe denaturation of the proteins, especially on a
membranes having a lower cut-off and adjusting the pH conditions, pilot plant scale, which is in accordance with previous findings
respectively (Waesche, Mueller, & Knauf, 2001). Moreover, the (Brueckner & Mieth, 1984). Furthermore, formation of firm aggre-
washing operations employed in the present study may account for gates during precipitation and separation is possible, resulting in
losses up to 17% of the precipitated proteins as shown in Fig. 4 for lower solubility. This assumption may be supported by the poor
IP-P1 and reported previously for IP-L2 (Pickardt et al., 2011). solubility of the globulin isolate even at low denaturation (Pickardt
Consequently, for industrial implementation, efficient water recy- et al., 2011). The low solubility of the globulin isolates might also be
cling is required in order to recover these proteins. ascribed to the intrinsic protein properties. These include the
Minor protein loss may also be ascribed to the adsorption pro- higher proportion of hydrophobic amino acids compared to e.g. soy
cess (Fig. 4), but more to the washing of the batch than to protein proteins and the depletion of hydrophilic amino acids during the
binding to the adsorber resin (Weisz et al., 2010). isolation process (Brueckner & Mieth, 1984). Low protein solubil-
ities of about 24e30% were also observed for globulin isolates
3.4. Functional properties of protein isolates (Kabirullah & Wills, 1982; Minones Conde et al., 2005; Yoshie-Stark
et al., 2008), whereas much higher values of 50% to > 80% have been
To assess the suitability of the derived products for different reported elsewhere (Gonza lez-Pe rez et al., 2002; Pawar et al., 2001;
food applications, selected functional properties were determined. Salgado et al., 2011). The different findings are mainly due to the
Practical assays were applied which have been adapted to certain different methods used for the protein preparation and solubility
food systems. Fig. 3 summarises the functional properties of the measurement, including different pH values which greatly affect
protein isolates and concentrates that were produced on a labora- protein solubility (Pickardt et al., 2009).
tory and pilot plant scale. The protein solubility was deemed to be a Low protein solubility is not necessarily disadvantageous for
decisive parameter because it is associated with other functional many applications of protein isolates. For applications requiring
properties (Gonza lez-Perez & Vereijken, 2007; Kabirullah & Wills, high protein solubility, this feature may be modified by hydrolysis
1981, 1982; Karayannidou et al., 2007; Salgado et al., 2011). The or other measures (Kabirullah & Wills, 1981, 1982; Karayannidou
emulsifying and foaming properties were also considered to be et al., 2007; Minones Conde et al., 2005).
promising for the sunflower protein isolates, whereas the gelling
properties were less notable, as revealed in preliminary tests and 3.4.2. Emulsifying capacity
literature data (Gonza lez-Perez & Vereijken, 2007). In our pre- The emulsifying capacities varied between 180e285 and
liminary studies, different sunflower protein isolates exhibited 510e640 mL/g for IP and UF products, respectively, with signifi-
weak gel strengths being far below that of soy protein isolates, and cantly lower values for the isolates obtained on a pilot plant scale.
high protein concentrations (>12%) and high gelation temperatures Values around 200 mL/g represent a moderate emulsifying capac-
(>95  C) were required for gel formation. ity, while ~600 mL/g is quite high compared to 800e1000 mL/g for
sodium caseinate (data not shown) which serves as a reference.
3.4.1. Protein solubility Furthermore, the following values measured for soy protein isolates
Protein solubilities at pH 7 were generally low for the precipi- in our laboratory may be taken as references: 645 mL/g for Supro
tated proteins with values in the 7.3e8.4% range for IP-P and in the 500E; 670 mL/g for Supro EX33, both from Dupont Protein Tech-
10.3e13.0% range for IP-L. The pilot plant proteins revealed slightly, nologies; 715 mL/g ProFam 464, ADM. Therefore, the emulsifying
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 217

activities of the albumin concentrates (UF) can be considered as impurities which might both affect functionality (Gonz rez
alez-Pe
very high, while being fair for the helianthinin isolates (IP). et al., 2005a). In contrast, Salgado, Ortiz,et al. (2012) did not find
Our results are in agreement with literature reports confirming an influence of different phenolic contents on the foaming prop-
the ability of both helianthinin and albumin preparations to form erties of sunflower protein concentrates.
stable emulsions (Gonza lez-Perez et al., 2005). However, literature Our results are in agreement with previous reports demon-
data on the functional properties of sunflower proteins are con- strating lower foaming activities but higher foam stabilities for
tradictory, probably due to the diversity of methods and pre- helianthinin compared to sunflower albumins (Gonz rez
alez-Pe
treatments used and the large variety of products investigated. et al., 2005a). The high foaming activities in the range of egg
Due to the different methods used, quantitative comparison with white and the high foam stability of the precipitated proteins are in
other studies is impossible. Using similar evaluation parameters, accordance with earlier reports (Kabirullah & Wills, 1982; Raymond
Canella et al. (1985) also described slightly higher emulsifying ca- et al., 1985) which also recommended sunflower proteins as
pacities of 172 mL/g for an albumin-enriched isolate compared to whipping agents. Raymond et al. (1985) reported the foaming
116 mL/g for a globulin isolate. Comparing our results to other properties of a sunflower protein preparation to be superior to that
proteins produced in our laboratory using essentially the same of soy proteins. However, quantitative comparison is virtually
methods, emulsifying capacities of 400 and 693 mL/g have been impossible due to the different methods and products, as
reported for rapeseed protein isolates obtained by precipitation mentioned above (Section 3.4.2). Our preliminary studies showed
and ultrafiltration, respectively. Values of 495 and 800 mL/g were that the foaming properties are highly dependent on the protein
measured for whole egg and egg white, respectively (Yoshie-Stark concentrations, whipping times and the foaming method. Using
et al., 2008). The emulsifying capacities of lupin protein isolates similar methodology, slightly lower foaming activities of ~1000%
were reported to range from 385 to 570 mL/g (D'Agostina et al., and stabilities >80% were reported for lupin proteins (D'Agostina
2006), thus, in a range similar to our samples. et al., 2006), which are known to have excellent foaming properties.
The generally higher emulsifying activity of the albumin con- In contrast to the emulsifying properties, the foaming activities
centrates (UF) compared to the helianthinin isolates (IP) may be did not show a clear correlation to the protein solubilities of the
ascribed to both the solubility and molecular size, with faster protein isolates (Fig. 3). This is in agreement with previous reports
diffusion of smaller molecules to the fluid interfaces (Minones (Gonz rez et al., 2005a) assuming solubility to be a less
alez-Pe
Conde et al., 2005) and greater molecular flexibility (Karayannidou important factor as only a small proportion of the protein might be
et al., 2007). As shown in Fig. 3, there is a clear correlation with incorporated in the foam. Higher precipitation efficiency at
the solubility of both the different types of proteins, which in turn is increased salt concentration and upon upscaling resulting in firmer
dependent on the intrinsic properties including the different amino aggregates that are less prone to dissolution during whipping may
acid composition as discussed above (Section 3.4.1). The differences further impair the foaming activity. While the foaming activity may
between the precipitated isolates on the different production scales be enhanced by the greater molecular flexibility of the smaller al-
seem to be correlated to the protein solubility which also was lower bumins compared to the larger globulins, albumins were found to
on a pilot plant scale (Fig. 3). This correlation may be ascribed to the be unable to stabilise the foams, probably due to their smaller
fact that the functional behaviour is governed by the soluble portion molecular size (Minones Conde et al., 2005). Higher molecular size
of the protein being able to unfold and interact at the oilewater- and increased viscosity of IP dispersions due to their lower solu-
interface, while the insoluble parts scarcely contribute to this bility may be another reason for the better stabilisation of foams by
functionality (Gonza lez-Perez et al., 2005). Moreover, denaturation globular proteins. This concurs with the findings of Kabirullah and
of protein isolates at pilot plant scale due to harsher process con- Wills (1981) who found hydrolysed sunflower protein isolates
ditions as discussed above (Section 3.4.1) may decrease emulsifying having greatly enhanced solubility to have better foam volume but
capacities, since the ability of the protein molecules to unfold and decreased stability.
align to the oilewater interface is lowered. Furthermore, formation
of firm aggregates during precipitation and separation may occur on 3.4.4. Comparison of different NaCl levels and different production
a pilot plant scale, thus limiting the formation of protein films at the scales
oilewater interfaces. Consequently, process conditions need to be Comparing the different NaCl levels, there are no clear-cut ef-
thoroughly adjusted for further process optimisation, with focus on fects on solubility and emulsifying properties, while the foaming
avoiding very low pH values for precipitation. activities were slightly higher for the lower salt level. Regarding the
different production scales, the protein solubilities and foaming
3.4.3. Foaming activity activities of the isolates from the pilot plant scale were only slightly
The foaming capacity was very high for all precipitated samples lower. However, the emulsifying capacities were markedly lower. It
with overrun values between 1063 and 1299% with higher values at should be noted that slightly different processing conditions, in
the lower salt level, while differences between the production scales particular for the precipitation step, were used on the different
were less pronounced. Significant differences were found comparing production scales. On a pilot plant scale, the acid used for fast pH
IP-P1 and IP-P2. UF concentrates provided even higher foaming ca- adjustment of high volumes was of higher concentration (3 mol/L
pacities of 1407e1500%, which were significantly higher as compared HCl), probably causing local pH dips as well as a generally slightly
to the IP of the respective salt concentration. However, these foams lower pH level. Furthermore, a pasteurisation step was performed
rapidly collapsed, while IP foams were very stable keeping 85e100% on pilot plant scale, which was necessary to ensure product safety
of their initial volume over 1 h (data not shown). This compares to for potential food applications. Therefore, greater denaturation of
1100e1600% activity for egg white foam with a foam stability of 90% the proteins on a pilot plant scale might have occurred, which is
(data not shown), which may serve as a reference. likely to affect the functional properties even though protein sol-
Slightly higher values were generally found for the lower salt ubility at pH 7 was only slightly decreased (Pickardt et al., 2011).
level. These do not seem to be clearly related to solubility values, The inferior functionality of the protein isolates produced on a pilot
which will be further discussed below. Comparing the composition plant scale depicted in Fig. 3 is mainly ascribed to such process
of the products, only small differences can be found which cannot variations. Consequently, in order to avoid protein denaturation,
serve as an explanation either. Only the significantly darker colour the precipitation conditions must be carefully controlled in further
may indicate some bonding of phenolic compounds or further scale-up work.
218 C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219

4. Conclusions Brueckner, J., & Mieth, G. (1984). Sunflower seeds and their processed products. Part
3. Functional properties and application of products enriched by protein.
Nahrung, 28(9), 933e953.
Sunflower protein isolates of light colour were produced from Canella, M., Castriotta, G., Bernardi, A., & Boni, R. (1985). Functional properties of
de-oiled press cake by the novel process that combines mild-acidic individual sunflower albumin and globulin. LWT e Food Science and Technology,
protein extraction at pH 6 with adsorptive removal of phenolic 18, 288e292.
Cater, C. M., Gheyasuddin, S., & Mattil, K. F. (1972). Effect of chlorogenic, quinic, and
compounds from the protein crude extracts. The adsorption of caffeic acids on the solubility and color of protein isolates, especially from
phenolic compounds from the protein extracts improved the colour sunflower seed. Cereal Chemistry, 49, 508e514.
of the protein isolates more effectively than hitherto proposed pre- D'Agostina, A., Antonioni, C., Resta, D., Arnoldi, A., Bez, J., Knauf, U., et al. (2006).
Optimization of a pilot-scale process for producing lupin protein isolates with
extraction processes confirming the general feasibility of the pro- valuable technological properties and minimum thermal damage. Journal of
cess. Despite the slight processing differences which occurred Agricultural and Food Chemistry, 54, 92e98.
during scale-up to pilot plant scale, the overall properties of the Davin, A., Berot, S., & Petit, L. (1983). Preparation of sunflower isolates at a pilot-
plant scale. Qualitas Plantarum e Plant Foods for Human Nutrition, 33, 221e226.
protein isolates were comparable to those produced on a laboratory DGF. (2011). Determination of fat content by the Caviezel Method (Rapid Method)
scale. This demonstrates the feasibility of scaling up the process. C-III 19(00). In H.-J. Fiebig, & Deutsche Gesellschaft für Fettwissenschaft e.V.,
However, process times and process conditions must be carefully Münster (Eds.), DGF-Einheitsmethoden Deutsche Einheitsmethoden zur Untersu-
chung von Fetten, Fettprodukten, Tensiden und verwandten Stoffen (2nd ed.).
controlled in order to avoid extensive denaturation. Further pilot Stuttgart: WVG.
plant trials are needed to carry out precise protein quantification Gonz alez-Perez, S., van Koningsveld, G. A., Vereijken, J. M., Merck, K. B., Gruppen, H.,
and an economic assessment. Additionally, the simultaneous re- & Voragen, A. G. J. (2005). Emulsion properties of sunflower (Helianthus annuus)
proteins. Journal of Agricultural and Food Chemistry, 53, 2261e2267.
covery of polyphenols may improve the economic viability of the
Gonz alez-Perez, S., Merck, K. B., Vereijken, J. M., van Koningsveld, G. A., Gruppen, H.,
overall process. & Voragen, A. G. J. (2002). Isolation and characterization of undenatured
Comparing the different NaCl levels, the higher salt concentra- chlorogenic acid free sunflower (Helianthus annuus) proteins. Journal of Agri-
tion was confirmed to enhance protein recovery by far, while the cultural and Food Chemistry, 50, 1713e1719.
Gonz alez-Perez, S., & Vereijken, J. M. (2007). Sunflower proteins: overview of their
lower level was slightly better with regard to the quality, especially physicochemical, structural and functional properties. Journal of the Science of
the colour, of the protein isolates. Therefore, higher salt concen- Food and Agriculture, 87, 2173e2191.
trations should be applied in further scale-up optimisation, while Gonz alez-Perez, S., Vereijken, J. M., van Koningsveld, G. A., Gruppen, H., &
Voragen, A. G. J. (2005a). Formation and stability of foams made with sunflower
the amount of resin and time of adsorption must be customised in (Helianthus annuus) proteins. Journal of Agricultural and Food Chemistry, 53,
order to compensate the impaired adsorption efficiency in higher 6469e6476.
salt concentrations. Moreover, extraction at an intermediate salt Gonz alez-Perez, S., Vereijken, J. M., van Koningsveld, G. A., Gruppen, H., &
Voragen, A. G. J. (2005b). Physicochemical properties of 2S albumins and the
level appears to be a promising approach for optimising the protein corresponding protein isolate from sunflower (Helianthus annuus). Journal of
yield whilst maintaining high adsorption capacity. Food Science, 70, C98eC103.
Regarding product performance, our results show that sun- Kabirullah, M., & Wills, R. B. H. (1981). Functional properties of sunflower protein
following partial hydrolysis with proteases. Lebensmittel-Wissenschaft und
flower protein isolates recovered by the novel process have -Technologie, 14, 232e236.
promising functional properties for their application in a variety of Kabirullah, M., & Wills, R. B. H. (1982). Functional properties of acetylated and
food products, in particular as emulsifiers and whipping agents. In succinylated sunflower protein isolate. Journal of Food Technology, 17,
235e249.
preliminary studies, such protein ingredients have been success-
Karayannidou, A., Makri, E., Papalamprou, E., Doxastakis, G., Vaintraub, I., Lapteva, N.,
fully used in the production of vegan (egg-free) mayonnaise and et al. (2007). Limited proteolysis as a tool for the improvement of the func-
foamed dessert products as well as for the replacement of egg tionality of sunflower (Helianthus annuus L.) protein isolates produced by seeds
white in marshmallow type confectionery products and sponge or industrial by-products (solvent cake). Food Chemistry, 104, 1728e1733.
Kausar, T., Ali, S., Javed, M. A., & Javed, M. A. (2004). Sunflower seed: a potential
cake formulations. source of food and feed products. Proceedings of the Pakistan Academy of Sci-
ences, 41, 49e54.
Kausar, T., Ali, S., Javed, M. A., & Khan, A. D. (2002). Effect of heat processing con-
Acknowledgements ditions on the protein quality of sunflower meal. Proceedings of the Pakistan
Academy of Sciences, 39, 83e87.

The authors thank Dr. Rass, Teutoburger Olmühle GmbH & Co. Kortt, A. A., & Caldwell, J. B. (1990). Sunflower 11 S globulin, susceptibility to pro-
teolytic cleavage of the subunits of native helianthinin during isolation: HPLC
KG (Ibbenbüren, Germany) for the provision of the sunflower press fractionation of the subunits. Phytochemistry, 29, 1389e1396.
cake. We are grateful to Prof. Dr. Endreb, Herbstreith & Fox KG Minones Conde, J., Yust Escobar, M.d. M., Pedroche Jimenez, J. J., Millan
(Neuenbürg, Germany) for his gift of adsorbent resin to perform the Rodriguez, F., & Rodriguez Patino, J. M. (2005). Effect of enzymatic treatment of
extracted sunflower proteins on solubility, amino acid composition, and surface
pilot plant experiments, and to Dr. Georg Weisz (Hohenheim Uni- activity. Journal of Agricultural and Food Chemistry, 53, 8038e8045.
versity) for providing the protocol for the adsorption experiments. Morr, C. V., German, B., Kinsella, J. E., Regenstein, J. M., Van Buren, J. P., Kilara, A.,
The authors acknowledge the valuable assistance of Michael Schott et al. (1985). Collaborative study to develop a standardized food protein solu-
bility procedure. Journal of Food Science, 50, 1715e1718.
and Michael Frankl with the pilot plant experiments and thank
Pandya, M. J., Sessions, R. B., Williams, P. B., Dempsey, C. E., Tatham, A. S.,
Sigrid Bergmann, Sigrid Gruppe, Elfriede Bischof, Maria Hillreiner, Shewry, P. R., et al. (2000). Structural characterization of a methionine-rich,
Isaac Tenllado and Thomas Hager for their help with the physical emulsifying protein from sunflower seed. Proteins: Structure, Function, and
and chemical analyses and laboratory sample preparation. Bioinformatics, 38, 341e349.
Pawar, V. D., Patil, J. N., Sakhale, B. K., & Agarkar, B. S. (2001). Studies on selected
This research project was supported by the German Ministry of functional properties of defatted sunflower meal and its high protein products.
Economics and Technology (via AiF (Arbeitsgemeinschaft indus- Journal of Food Science and Technology, 38, 47e51.
trieller Forschungsvereinigungen “Otto von Guericke” e.V.)) and the Pickardt, C., Hager, T., Eisner, P., Carle, R., & Kammerer, D. R. (2011). Isoelectric
protein precipitation from mild-acidic extracts of de-oiled sunflower (Helianthus
FEI (Forschungskreis der Ern€ ahrungsindustrie e.V., Bonn), Project annuus L.) press cake. European Food Research and Technology, 233, 31e44.
AiF 14449 N. Pickardt, C., Neidhart, S., Griesbach, C., Dube, M., Knauf, U., Kammerer, D. R., et al.
(2009). Optimisation of mild-acidic protein extraction from defatted sunflower
(Helianthus annuus L.) meal. Food Hydrocolloids, 23, 1966e1973.
References Prasad, T. D. (1990). Proteins of the phenolic extracted sunflower meal: 1. Simple
method for removal of polyphenolic components and characteristics of salt-
AOAC. (2005a). Method 923.03. Ash of flour. In G. W. Latimer, & W. Horwitz (Eds.), soluble proteins. Lebensmittel-Wissenschaft und -Technologie, 23, 229e235.
Official methods of analysis of AOAC international (18th ed.). Gaithersburg: AOAC Raymond, J., Inquello, V., & Azanza, J. L. (1991). The seed proteins of sunflower:
International. comparative studies of cultivars. Phytochemistry, 30, 2849e2856.
AOAC. (2005b). Method 968.06. Protein (crude) in animal feed. In G. W. Latimer, & Raymond, J., Rakariyatham, N., & Azanza, J. L. (1985). Functional properties of a new
W. Horwitz (Eds.), Official methods of analysis of AOAC international (18th ed.). protein isolate from sunflower oil cake. Lebensmittel-Wissenschaft und -Tech-
Gaithersburg: AOAC International. nologie, 18, 256e263.
C. Pickardt et al. / Food Hydrocolloids 44 (2015) 208e219 219

Saeed, M., & Cheryan, M. (1988). Sunflower protein concentrates and isolates low in Thiyam, U., Sto€ ckmann, H., Zum Felde, T., & Schwarz, K. (2006). Antioxidative effect
polyphenols and phytate. Journal of Food Science, 53, 1127e1131. of main sinapic acid derivatives from rapeseed and mustard oil by-products.
Salgado, P. R., Drago, S. R., Molina Ortiz, S. E., Petruccelli, S., Andrich, O., European Journal of Lipid Science and Technology, 108, 239e248.
Gonzalez, R. J., et al. (2012). Production and characterization of sunflower Waesche, A., Mueller, K., & Knauf, U. (2001). New processing of lupin protein iso-
(Helianthus annuus L.) protein-enriched products obtained at pilot plant scale. lates and functional properties. Nahrung, 45, 393e395.
LWT e Food Science and Technology, 45, 65e72. Weisz, G. M., Carle, R., & Kammerer, D. R. (2013). Sustainable sunflower processing
Salgado, P. R., Molina Ortiz, S. E., Petruccelli, S., & Mauri, A. N. (2011). Sunflower d II. Recovery of phenolic compounds as a by-product of sunflower protein
protein concentrates and isolates prepared from oil cakes have high water extraction. Innovative Food Science and Emerging Technologies, 17, 169e179.
solubility and antioxidant capacity. Journal of the American Oil Chemists' Society, Weisz, G. M., Kammerer, D. R., & Carle, R. (2009). Identification and quantification of
88, 351e360. phenolic compounds from sunflower (Helianthus annuus L.) kernels and shells
Salgado, P. R., Ortiz, S. E. M., Petruccelli, S., & Mauri, A. N. (2012). Functional food by HPLC-DAD/ESI-MSn. Food Chemistry, 115, 758e765.
ingredients based on sunflower protein concentrates naturally enriched with Weisz, G. M., Schneider, L., Schweiggert, U., Kammerer, D. R., & Carle, R. (2010).
antioxidant phenolic compounds. JAOCS, Journal of the American Oil Chemists' Sustainable sunflower processing d I. Development of a process for the
Society, 89, 825e836. adsorptive decolorization of sunflower [Helianthus annuus L.] protein extracts.
Schieber, A., Stintzing, F. C., & Carle, R. (2001). By-products of plant food processing Innovative Food Science and Emerging Technologies, 11, 733e741.
as a source of functional compounds e recent developments. Trends in Food Xu, L., & Diosady, L. L. (2002). Removal of phenolic compounds in the production of
Science and Technology, 12, 401e413. high-quality canola protein isolates. Food Research International, 35, 23e30.
Shchekoldina, T., & Aider, M. (2012). Production of low chlorogenic and caffeic acid Yoshie-Stark, Y., Wada, Y., & Waesche, A. (2008). Chemical composition, functional
containing sunflower meal protein isolate and its use in functional wheat bread properties, and bioactivities of rapeseed protein isolates. Food Chemistry, 107,
making. Journal of Food Science and Technology, 49, 1e13. 32e39.
Taha, F. S., Abbasy, M., El-Nockrashy, A. S., & Shoeb, Z. E. (1981). Countercurrent Zilic, S., Barac, M., Pesic, M., Crevar, M., Stanojevic, S., Nisavic, A., et al. (2010).
extraction-isoelectric precipitation of sunflower seed protein isolates. Journal of Characterization of sunflower seed and kernel proteins. Helia, 33, 103e114.
the Science of Food and Agriculture, 32, 166e174.

Вам также может понравиться