Thermal stability of reconstituted milk protein concentrates: Effect of partial calcium depletion during membrane filtration

Accepted Manuscript Thermal stability of reconstituted milk protein concentrates: Effect of partial calcium depletion during membrane filtration

H. Eshpari, R. Jimenez-Flores, P.S. Tong, M. Corredig PII: DOI: Reference:

S0963-9969(17)30398-8 doi: 10.1016/j.foodres.2017.07.058 FRIN 6855

To appear in:

Food Research International

Received date: Revised date: Accepted date:

18 March 2017 23 July 2017 26 July 2017

Please cite this article as: H. Eshpari, R. Jimenez-Flores, P.S. Tong, M. Corredig , Thermal stability of reconstituted milk protein concentrates: Effect of partial calcium depletion during membrane filtration, Food Research International (2017), doi: 10.1016/ j.foodres.2017.07.058

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Food Research International

T

Title:

CR

IP

Thermal Stability of Reconstituted Milk Protein Concentrates: Effect of Partial Calcium Depletion during Membrane Filtration

US

H. Eshpari*1, R. Jimenez-Flores†2, P.S. Tong †2, M. Corredig*2

*

Department of Food Science, University of Guelph, Guelph, Ontario N1G 2W1, Canada Dairy Products Technology Center, California Polytechnic State University, San Luis Obispo 93407

1

ED PT

CE

Keywords: Milk protein concentrate Casein Calcium depletion Heat stability Non-sedimentable protein Serum composition

M

AN

†

2

AC

H. Eshpari e-mail: [email protected] Current address: Tillamook Cheese, 6830 SW Atlanta St., Tigard, OR 97223 Corresponding author: M. Corredig e-mail: [email protected] R. Jimenez-Flores e-mail: [email protected] P.S. Tong e-mail: [email protected] 1

ACCEPTED MANUSCRIPT Abstract Milk protein concentrate (MPC) powders are increasingly utilized in manufacturing of protein fortified beverages. Thermal stability of the protein dispersions is of significant importance in such applications. It is known that a decrease in pH can induce partial dissociation of casein micelles

T

and modify the natural equilibrium of calcium and phosphate between the micelles and the serum

IP

phase. The presence of soluble casein may improve the rehydration properties of MPC powders,

CR

and may impact their thermal stability. The objective of this work was to investigate the effects of

US

partial acidification of milk prior to ultrafiltration on the heat stability of reconstituted MPC dispersions. Milk protein concentrate powders were prepared from skim milk acidified to pH 6.0

AN

by addition of glucono-δ-lactone, and then concentrated using ultrafiltration (UF) and diafiltration (DF). The heat stability of the reconstituted MPC dispersions was studied, by determining heat

M

coagulation time, particle size, turbidity, viscosity, soluble and colloidal calcium and phosphate,

ED

and non-sedimentable casein both before and after heating at 120°C. Reconstituted MPC powders

PT

made with partially acidified skim milk contained lower soluble calcium and phosphate and exhibited very poor thermal stability compared to MPC powders made with skim milk at its natural

CE

pH. The thermal stability of the acidified MPC dispersions was not only recovered by restoration of pH and the serum composition through dialysis against skim milk, but it was improved

AC

compared to control MPC dispersions. All dialyzed samples had comparable pH, protein content and calcium and phosphate concentration, but the structure of the casein micelles was altered, causing differences in the type of soluble aggregates. It was concluded that the integrity of the casein micelles and the amount of dissociated, non sedimentable caseins play a major role in determining the thermal stability of the MPC dispersions.

2

ACCEPTED MANUSCRIPT 1. Introduction Thermal stability is an important property of milk proteins, as dairy formulations must withstand heating temperatures sufficient to decrease the risk of pathogens contamination, without exhibiting any physical instability, such as phase separation, gelation or visible coagulation. It is

T

known that the heat induced destabilization of milk can be promoted by several changes to the

IP

casein micelles and their surrounding serum environment caused by heating. Changes in pH,

CR

calcium and phosphate deposition on the micelles, dephosphorylation, dissociation and hydrolysis,

US

whey proteins-caseins interactions and formation of protein aggregates through hydrophobic and covalent disulphide bonds have been reported (Fox, 1981).

AN

Whey proteins play a major role in heat induced destabilization of milk; these proteins are non-phosphorylated, heat- and calcium sensitive, and can be involved in thiol-disulfide

M

interchanges that result in covalent aggregates (McKenzie, 1971; Fox & Mulvihill, 1982). Heating

ED

milk at temperatures above 70°C leads to irreversible denaturation of whey proteins (Swaisgood,

PT

1982). The denatured whey proteins will covalently bind to κ-casein on the surface of casein micelles or end up in the serum phase in the form of non-sedimentable complexes, depending on

CE

pH and serum composition of the milk (Guyomarc'h, Law, & Dalgleish, 2003; Donato et al. 2007; Li, Dalgleish, & Corredig, 2015). A high concentration of non sedimentable caseins may also

AC

affect the composition and stability of the heat induced colloidal particles; however, this hypothesis has never been tested. The mineral balance between the colloidal and soluble phases of milk is crucial to its thermal stability. It has been reported that during heating of milk, serum phosphate and calcium decrease as the ions become insoluble, causing an overall decrease of the negative charges on the micelles and a decrease in charge repulsion (Singh & Creamer, 1991; Anema, 2009).

3

ACCEPTED MANUSCRIPT Lactose also plays an important role in thermal stability of milk, as its heat induced oxidation and degradation to organic acids can cause a pH decrease in the serum phase (Dalgleish, 1989). The presence of lactose during heating also seem to reduce the calcium-binding capacity of the caseins because of reduced accessibility of binding sites due to Maillard reactions (Pappas &

T

Rothwell, 1991). This will affect the colloidal stability of the casein micelles.

IP

Milk protein concentrates (MPCs) can vary in protein contents from 35% to 85% protein (on

CR

dry basis) and are manufactured by concentrating skim milk, using processes such as ultrafiltration (UF), diafiltration (DF), evaporation, and spray-drying. Numerous studies are available on the

US

rehydration and functional properties of milk protein concentrates (MPC); however, there is

AN

limited knowledge on the heat stability of rehydrated MPC powders, especially as a function of changes in the processing history or rehydration conditions. This is of particular interest, as new

M

practices are showing the potential of a partial modification of milk protein concentrates to tailor

al., 2014; Uluko et al., 2016).

ED

their processing functionality and create value added ingredients (Marella et al., 2015; Eshpari et

PT

As in milk, also in reconstituted MPC, the equilibrium between soluble and insoluble calcium

CE

is known to be critical to the stability of casein micelles. For example, Crowley et al. (2014) demonstrated that the heat stability of rehydrated MPC powders could be restored after re-

AC

establishment of the serum composition to that of the original milk. Studies have also been conducted in milk protein concentrates depleted of whey proteins, such as micellar casein concentrates. Changes in the mineral equilibrium and partial dissociation of the casein micelle occurring in micellar casein concentrate during heating at high temperatures play an important role in heat induced aggregation and coagulation of the dispersions (Sauer & Moraru, 2012). Increasing the pH or decreasing the heating temperature, improve the stability of the dispersions (Sauer &

4

ACCEPTED MANUSCRIPT Moraru, 2012). It is known that during concentration of milk by membrane filtration, a higher level of soluble caseins is obtained, and the concentration and type of heat induced aggregates, after heating at 80°C for 15 min changes depending on the extent of the concentration and the heat treatment (Li & Corredig, 2014).

T

It is known that calcium chelators can increase the heat stability of milk solutions by

IP

decreasing the concentration of free calcium ions in the serum phase, hence reducing the calcium-

CR

induced protein aggregation (Leviton & Pallansch, 1962; Augustin & Clarke, 1990). However, depending on the concentration added, calcium chelators can also chelate colloidal calcium

US

phosphate from the casein micelle to an extent that their structural integrity is lost (Griffin, Lyster,

AN

& Price, 1988; Augustin & Clarke, 1990).

de Kort et al. (2012) demonstrated that calcium chelators have different influences on the

M

micellar structure and physico-chemical properties of micellar casein concentrate solutions and

ED

can induce swelling, dissociation, and/or aggregation of the casein micelles. Therefore, modification of viscosity, turbidity, and heat stability of the casein-based concentrates could be

PT

achieved by choosing the appropriate type and concentration of calcium chelator. Previous work

CE

on partly acidified samples (Eshpari et al., 2014) have shown the potential of varying the heat stability of the reconstituted powders. However, a detailed understanding of the physical chemical

AC

changes occurring during heating treatment was not reported. The objective of this study was to investigate the impact of heat treatment at 120 °C on the colloidal properties of casein micelles in reconstituted milk protein concentrates. In particular, this research focused on the effect of partial acidification of milk to pH 6.0 by addition of glucono-δlactone prior to ultrafiltration on the thermal stability of the resulting MPC powders, after rehydration at equal 3.2% (w/w) protein. It was hypothesized that such decrease in pH, would

5

ACCEPTED MANUSCRIPT cause a decrease in calcium and an increase in soluble caseins, resulting in improved thermal stability of the MPC dispersions. To determine the effect of compositional modifications in the serum phase of the acidified MPC samples on their heat stability, the MPC dispersions were dialyzed against skim milk. As MPC dispersions are increasingly utilized in formulating high

T

protein beverages, it is important to fully understand their impact on the heat stability of the

CR

IP

mixtures and the potential of modifying their processing history to tailor a specific functionality.

US

2. Materials and Methods 2.1. Preparation of milk protein concentrates

AN

Pasteurized skim milk was obtained from Producer’s Dairy Foods Inc. (Fresno, CA). Analytical grade reagents were from Sigma-Aldrich Chemical Ltd. (St. Louis, MO). Glucono-δ-

M

lactone (GDL) was purchased from Roquette America, Inc. (Geneva, IL). Ultrapure water (Milli-

ED

Q Ultrapure Water Purification Systems, Billerica, MA) was used to prepare all the solutions.

PT

Milk protein concentrate (MPC) powders were manufactured in duplicate, either by ultrafiltration (65% protein, MPC 65) or by ultrafiltration followed by diafiltration (80% protein,

CE

MPC80), using pasteurized skim milk as previously described in detail (Eshpari et al., 2014). Controls were prepared at the native milk pH (~ pH 6.6), or GDL was added (at 3.25 g/L) to reach

AC

pH 6.0 before starting membrane filtration. The MPC65 and MPC80 powders were manufactured in the pilot plant of Dairy Products Technology Center at California Polytechnic State University (San Luis Obispo) with an R12 cross-flow membrane pilot-plant unit (Niro Inc., Hudson, WI) equipped with dual 10-kDa cut-off, spiral-wound, polyethersulfone membranes (Snyder Filtration, Vacaville, CA). The liquid MPC was spray-dried with a pilot Niro Filtermat Spray Dryer (Niro Inc.) to approximately 3.5% moisture, and the obtained MPC powders were immediately collected

6

ACCEPTED MANUSCRIPT and sealed in airtight bags for further analysis. Total protein present in the powders was determined by Kjeldahl. Ash content was determined by ignition at 550 °C in an electric muffle furnace. Fat content was determined by the Mojonnier method and free moisture content by oven-drying method.

T

2.2. Powder rehydration

IP

MPC powders were reconstituted in Ultrapure water (Milli-Q Ultrapure Water Purification

CR

Systems, Billerica, MA) at 25 °C to a final protein concentration of 3.2% (w/w) using a household kitchen blender for 6 min at high speed. To ensure complete solubilization of the powder particles,

US

all MPC dispersions were subsequently subjected to homogenization at 150 bar for two passes

AN

using Emulsiflex C5, Avestin (Ottawa, Ontario, Canada). Each sample was then divided into two equal portions; one portion was stored in a tightly sealed container at 4°C overnight. The other

M

portion of the rehydrated samples (prepared as described above) was dialyzed against skim milk

ED

at 4°C also overnight, to allow enough time for full rehydration and obtain samples with similar rehydration history before and after dialysis. Dialysis was carried out using an 8000 Da cut-off

PT

membrane (Spectra/Por, Spectrum Laboratories, Inc., Rancho Dominguez, CA, USA) to restore,

CE

as much as possible, the original milk serum composition and to compare the heat coagulation behavior of the casein micelles between treatments.

AC

The control samples were named UFC and DFC before and UFC-D and DFC-D after dialysis, similarly the GDL treated samples were named UFG and DFG before and UFG-D and DFG-D after dialysis. 2.3. Evaluation of heat stability The heat coagulation time (HCT) of the MPC solutions (3.2% w/w protein) was determined following the method described by Davies & White (1960) with some modifications: a 3-mL

7

ACCEPTED MANUSCRIPT aliquot of each solution was transferred to a heat-resistant screw-cap test tube (internal diameter 10 mm, length 100 mm). The test tubes were fixed to a metal platform and immersed in an oil bath maintained at 120°C. The HCT was recorded as the time elapsed between immersing the samples in the hot oil bath and the onset of visible clots, hence a sample which required a longer time to

T

show visible coagulation was considered more heat stable and vice versa. Upon immersing the

IP

samples in the oil bath, a come up time of 30 s was necessary for the samples to reach 120 °C. The

CR

come up time was not counted in HCT. The heated samples were cooled down to 22°C using ice water before conducting any measurements. The HCT of the solutions was tested both before and

US

after dialysis against skim milk. Samples were also collected after 10 min of heating in the oil

AN

bath, a time that was arbitrarily chosen to better compare all treatments. This time was below the HCT for most samples. The pH of samples both before and after heating was measured using an

M

Accumet pH meter model 925 (Fisher Scientific). All analyses were carried out in duplicate with

ED

separate batches of reconstituted MPC samples.

2.4. Physical properties of heated MPC dispersions

PT

The apparent diameter of the casein micelles in reconstituted MPC dispersions and a skim

CE

milk sample as control was determined using a dynamic light scattering (DLS) instrument (Zetasizer, Nano-ZS, Malvern Instruments, Malvern, UK), after extensive dilution with permeate.

AC

Permeate was filtered using 0.2 mm nylon filters (Fisher Scientific) and samples were diluted about 1000 times right before analysis at 22°C. Transmission diffusing wave spectroscopy (DWS) was employed to determine the turbidity parameter (1/l*) of the dispersions (Alexander & Dalgleish, 2004). Samples were measured before and after heating at 120 °C. The measurements were carried out at 25 °C and the instrument was calibrated daily using a standard latex (260 nm diameter; Portland Duke Scientific, Palo Alto, CA,

8

ACCEPTED MANUSCRIPT USA). Each sample was poured into an optical glass cuvette of 5 mm path length (Hellma Canada Limited, Concord, ON, Canada) and was placed in a water bath to maintain the temperature. The sample was then subjected to a solid-state laser light illumination with a wavelength of 532 nm and a power of 350 mW (Coherent, Santa Clara, CA, USA). Scattered light intensity was collected

T

in transmission mode as previously described (Sandra et al., 2011). The data was analysed using

IP

DWSFit (Mediavention Inc., Guelph, ON, Canada). DWS was used to determine the photon

CR

transport mean free path l* which is related to the length scale over which the direction of the scattered light has been totally randomized. Turbidity was calculated as the inverse of the l*

US

parameter.

AN

The viscosity of all MPC dispersions as well as a skim milk control, both before heating and after heating at 120 °C for 10 min or until the onset of heat coagulation, was measured using an

M

advanced Rheometer AR 1000 (TA Instruments Ltd., New Castle, DE, USA) equipped with a

ED

peltier temperature controller. Samples were subjected to a steady flow test (shear rate ramp from 0 to 100 s−1), using a cone and plate geometry, with a set gap of 0.51 mm. The values are reported

PT

for 100 s−1. All measurements were conducted at 25 °C.

CE

2.5. Calcium and Phosphate

The concentration of total, soluble and diffusible (permeable) ions were measured using ion

AC

chromatography, with an Advanced Compact IC (Metrohm AG, Zurich, Switzerland), using a silica gel column (Metrosep C2 150/40 packed with 7 mm silica gel; Metrohm AG) at 30 °C, following the method described by Rahimi-Yazdi, Ferrer, & Corredig, (2010). The amount of calcium in the centrifugal supernatant of reconstituted powders (65,000 g for 1 h at 25°C) was referred to as soluble (non sedimentable) calcium, containing both protein-bound and protein-free calcium fractions of the sample. In a separate experiment, each sample were filtered using a

9

ACCEPTED MANUSCRIPT Prep/ScaleTM-TEF 1ft2 cartridge ultrafiltration unit (10 KDa cut-off Regenerated cellulose, Millipore, Ontario, Canada) to separate the protein free permeate. The calcium measured in the permeate was referred to as permeable calcium or non-protein bound soluble calcium. The amount of soluble and permeable phosphate was also measured in the MPC samples. As

T

for the calcium ions, the amount of phosphate in the centrifugal supernatant of the samples (65,000

IP

g for 1 h at 25°C) was referred to as soluble (non sedimentable) phosphate, containing both protein-

CR

bound and protein-free calcium fractions of the sample. Permeable phosphate was instead measured by filtering samples through a UF cartridge as described above.

For phosphate

US

measurements, 1 mL of each fraction was transferred to a 5 ml Pyrex test tube (Corning

AN

Incorporated Life Sciences , MA, USA) and dry-ashed by initial heating at 100°C overnight to dry the samples, followed by further heating at 500 °C for 6 hours to remove the organic matter. The

M

dry-ashing was conducted in an Isotemp muffle furnace (Fisher Scientific, MA, USA). The

ED

obtained ashes were dissolved in 1 mL of 1 mol L-1 HNO3 to be injected into the HPLC. No dryashing was necessary in sample preparation for the permeable phosphate analysis, as the sample

PT

did not contain any bound proteins at this point. An anion column (Metrosep A Supp5-150/4.0,

CE

Metrohm) packed with 5 μm polyvinyl alcohol with quaternary ammonium groups was employed to measure the phosphate. Sodium hydrogen carbonate and sodium carbonate solutions were used

AC

to prepare mobile phase (1.0 mM sodium carbonate and 4mM sodium hydroxide). Samples were eluted at a flow rate of 0.5 mL min-1. The amount of colloidal calcium or phosphate was calculated by the difference between total and soluble calcium or phosphate. 2.6. Protein content of soluble fraction The total and soluble (in the supernatants after centrifugation 65,000 × g for 1 h at 25°C) protein contents of all samples were measured by the Dumas method (Leco FP-528) both before

10

ACCEPTED MANUSCRIPT and after heating the samples at 120 °C for 10 min. All analyses were carried out in duplicate with separate batches of reconstituted MPC samples. 2.7. SDS-PAGE of soluble fraction The centrifugal supernatant obtained by centrifugation at 65,000 × g for 1 h at 25°C was mixed

T

(1:1 ratio) with electrophoresis buffer, containing 1 M Tris HCl buffer pH 6.8, 75% glycerol, 10%

IP

SDS, 2-mercaptoethanol, 1% bromophenol blue. A skim milk sample (3.2% protein) before and

CR

after dialysis was used as control. The solution was heated and mixed at 95 °C for 5 min using a thermomixer (model 5436; Eppendorf, Hauppauge, NY, USA). Samples were cooled to room

US

temperature and 10 μl of each sample was then loaded onto the gels. SDS-PAGE was carried out

AN

in a vertical slab gel of 1.5 mm thickness with 15% acrylamide resolving gel and 4% stacking gel in a Bio-Rad mini-protean electrophoresis system (Bio-Rad Laboratories, Hercules, CA, USA) at

M

a constant voltage of 175 V. Gels were then treated with Coomassie blue R-250 for 45 min (Bio-

ED

Rad) and destained with 45% MilliQ water, 45% methanol, and 10% glacial acetic acid for 1 h, and then with the same solution diluted 1:1 in water overnight. The gels were scanned using a Bio-

PT

Rad Gel Doc EZ Imager (Bio-Rad Power Pac HC, Hercules, CA) equipped with Image Lab 3.0

CE

software (Bio-Rad Power Pac HC, Hercules, CA) and the protein bands were quantified by laser scanning densitometry.

AC

2.8. Statistical analysis

The sample treatments and analyses in the present study were performed in duplicate. Values were means of replicate determinations and the differences between the means of the treatments were compared by one-way ANOVA at a significance level of P < 0.05. The statistical analysis was conducted using the GLM command in Minitab (v.16.1, Minitab Inc., State College, PA).

11

ACCEPTED MANUSCRIPT Difference between the treatments were tested using Tukey’s honestly significant difference (HSD) intervals with α = 0.05.

3. Results and Discussion

T

3.1. Heat coagulation time and pH

IP

All MPC dispersions were reconstituted at 3.2% protein, to better compare between

CR

treatments. Preliminary research (Eshpari et al., 2014) suggested differences in the heat stability

US

of these dispersions, but no mechanistic understanding was derived, as dispersions were studied at different protein concentrations. After reconstitution in water, control samples UFC and DFC

AN

showed a significantly higher pH than skim milk (Table1). On the other hand, UFG showed a lower pH value compared to SM (pH 6.5), while the pH of the DFG dispersions was similar to that

M

of the original skim milk. As expected, after dialysis of the suspensions against skim milk all the

ED

treatments showed statistically equivalent pH values (Table 1). This was critical for comparison

PT

of samples, as it is well known that pH affects the thermal stability of milk proteins (Fox, 1981). Table 1 also presents the results of the heat coagulation experiments on the MPC dispersions,

CE

before and after dialysis. The heat coagulation time is defined as the time of the onset of visible instability during heating at 120°C. All MPC dispersions had a significantly shorter heat

AC

coagulation time compared to skim milk, when reconstituted at the same protein concentration. This behavior has been previously attributed to the reduction of serum milk constituents, such as, urea, lactose, and ionic phosphate, as these constituents show heat stability promoting potential (Metwalli & van Boekel 1996; Singh 2004). There was a statistically significant difference between control MPC (UFC) and GDL treated MPC (UFG), with the acidified samples showing the onset of heat coagulation immediately after reaching 120 °C. Furthermore, the heat stability of

12

ACCEPTED MANUSCRIPT DFC was lower than that of UFC, indicating that diafiltration further decreased the heat stability of MPC, despite DFC having a higher pH value than UFC. After dialysis, all MPC dispersions still showed a lower HCT compared to that of skim milk, despite a comparable serum composition amongst the dispersions. While in the case of UFC and

T

DFC, there were no significant differences in the heat coagulation time between before and after

IP

dialysis, in the case of the GDL treated samples, the heat stability significantly improved, and the

CR

coagulation times were longer than those of control UFC and DFC. This indicated that ion concentration and pH are critical, but that also the structure of the casein micelles and the ion-

US

protein interactions in the soluble phase are important factors in determining HCT.

AN

The pH of the heated MPC dispersions is also reported in Table 1. After heating for 10 min, the pH of MPC control (UFC and DFC) did not significantly change, while the GDL treated

M

samples showed a decreased pH of 6.26 and 6.6 for UFG and DFG, respectively. In other words,

ED

UFG showed a lower buffering capacity than DFG. The low pH was key to the low heat stability of these samples.

PT

All dialyzed MPC dispersions had a similar pH (about 6.6), and did not show a change after

CE

heating for 10 min, confirming that the buffering salts present in the soluble phase are key to maintain pH during heating. The serum composition (namely calcium, phosphate, and their

AC

different ionic form, as well as unsedimentable protein) is responsible for the significant differences observed in the colloidal stability of the casein micelles upon heating of these reconstituted samples. These findings confirmed previous reports (Crowley et al., 2014), who demonstrated that the composition of the serum phase is critical to the heat stability of milk.

13

ACCEPTED MANUSCRIPT 3.2. Heat-induced changes in particle size Table 2 summarizes the changes in the apparent hydrodynamic diameter of the casein micelles in the dispersions before and after heating for 10 min at 120 °C. Unheated control dispersions (UFC and DFC) showed a diameter of the casein micelles

T

similar to that of skim milk, both before and after dialysis. On the other hand, the reconstituted

IP

MPC samples treated with GDL showed a larger average diameter of about 197 and 207 nm, for

CR

UFG and DFG, respectively. These differences in particle size were maintained after dialysis against milk. It is important to note that it has been previously reported that a change in the mineral

US

equilibrium during partial acidification may cause a decrease of intra-micellar protein interactions,

AN

an increase in charge repulsion, and an increase in voluminosity of the casein micelles (Gaucheron, 2005). The present results also seemed to suggest that the change is due to protein rearrangements

M

within the reconstituted casein micelles, and not the soluble ion concentrations.

ED

After heating at 120°C for 10 min, skim milk samples showed an increase in the average diameter of the casein micelles. This increase is caused by the presence of whey protein aggregates

PT

associated with the casein micelles (see for example, Vasbinder & de Kruif, 2003). On the other

CE

hand, control UF and DF samples (UFC and DFC), before re-equilibration by dialysis, showed a significantly smaller size after heating compared to heated SM. This is likely due to the formation

AC

of soluble aggregates of whey proteins and caseins at a higher pH than that of skim milk; these aggregates are smaller in diameter than the casein micelles (Vasbinder & de Kruif. 2003). The increased release of casein molecules in the soluble phase during heating of milk can be responsible for the smaller micellar radius at the onset of heat coagulation (Donato, Alexander, & Dalgleish, 2007; Guyomarc'h et al. 2003). Denatured whey proteins, especially β-lactoglobulin, attached to the casein micelles has been recognized as a major contributing factor to the change in

14

ACCEPTED MANUSCRIPT the size of the casein micelles during heating (Anema & Li 2003). Anema (2007) demonstrated that heating milk at pH 6.4 leads to an increase of about 30 nm in radius of the micelle, coinciding with 85% of the whey proteins present in the colloidal phase, whereas at pH 7.1 the size is smaller than that of milk heated at normal pH of 6.7 with only 15% of the whey protein complexes

T

associated with the micelles. After dialysis, by restoring the pH and serum composition, the

IP

apparent diameter of the casein micelles was not significantly different from that of skim milk,

CR

regardless of the treatment.

In the suspensions prepared with MPC treated with GDL (UFG and DFG) aggregation

US

occurred during heating, with very short heat coagulation times (Table 1). After dialysis against

AN

milk, heating did not cause visible aggregation after 10 min, and all MPC samples showed very similar apparent diameter, confirming the importance of pH and buffering ions in controlling

M

protein aggregation.

ED

In conclusion, the thermal stability of MPC dispersions was affected by pH as well as other components present in the serum phase. The concentration of calcium and phosphate in the serum

PT

phase regulated the extent of dissociation of the casein micelles. Indeed, after restoration of the pH

CE

and serum composition by dialysis, all samples (at both UF and DF levels) exhibited improvement in thermal stability (as shown in Table 1) and similar values of apparent size of the casein micelles,

AC

although at a significantly lower level compared to skim milk. Furthermore, there was a substantial improvement in the heat stability for the GDL treated dispersions, which before dialysis showed fast coagulation. The processing history of the protein, the state of dissociation of the micelles, the amount of residual colloidal calcium and phosphate, lactose and pH were all critical factors in the heat stability of the reconstituted MPC samples.

15

ACCEPTED MANUSCRIPT 3.3 Turbidity The turbidity parameter (1/l*) for all MPC dispersions was measured using transmission DWS, and compared to that of skim milk, both before and after re-equilibration of the serum phase composition by dialysis (Table 3). The turbidity parameter is a function of the size and other

T

physical properties of the casein micelles as well as their interspatial correlation. All dispersions,

IP

before heating, showed a value of 1/l* statistically equivalent to that of skim milk. There were no

CR

differences in the optical properties of the casein micelles suspensions, regardless of the membrane filtration method or pH, before heating. Dialysis against milk also did not show a difference in the

US

value of 1/l* of the reconstituted samples. The turbidity parameter increased after 10 min heating

AN

at 120 °C, also for skim milk. The change for skim milk can be attributed to the increase in the apparent diameter of the casein micelles (Table 2). In the case of the control MPC (UFC and

M

DFC), the values of 1/l* were higher than those of corresponding unheated dispersions, but did not

ED

reach the same turbidity as the heated skim milk samples. In these samples, the apparent diameter of the casein micelles was lower in heated samples before dialysis, and similar to control after

PT

dialysis. After heating, the 1/l* values were higher for the dialyzed samples than for the same

CE

dispersions heated before dialysis.

The suspensions of reconstituted MPC originally acidified (UFG and DFG) showed a marked

AC

increase in the turbidity after heating, and a much lower value of 1/l* if dialyzed before heating (Table 3). These results are fully in line with the particle size data reported in Table 2. The differences between dispersions heated before or after dialysis pointed to the importance of the composition of the serum phase in the formation of heat induced aggregates. It is important to note that all dialyzed samples and skim milk exhibited similar values of hydrodynamic diameter of casein micelles after heating at 120 °C for 10 min (Table 2), but there were still differences in the

16

ACCEPTED MANUSCRIPT turbidity parameter between treatments. The increased turbidity of these dispersions after reequilibration and heating was caused by increased particle-particle interaction, as well as a change in the composition of the serum phase, namely, a change in the refractive index contrast. 3.4 Viscosity

T

As summarized in Table 3, all reconstituted MPC dispersions showed a lower viscosity compared

IP

to skim milk. After dialysis, the dispersions increased their viscosity, confirming that the changes

CR

were related to the composition of the serum phase. After heating at 120 °C for 10 min, only skim milk and UFC showed a lower viscosity value compared to the unheated dispersions; all the other

US

samples showed no statistically significant differences. After dialysis, the viscosity values of the

AN

heated samples were not different than those of the unheated counterparts. Preacidification with GDL during membrane filtration did not show an effect on the bulk viscosity compared to the

M

corresponding controls.

ED

3.5 Calcium and Phosphate concentrations before and after heating All reconstituted MPC dispersions had a lower amount of total calcium and phosphate,

PT

compared to skim milk, indicating a statistically significant loss of the minerals in the permeate

CE

during membrane filtration (Fig. 1A and 2A). In addition, both preacidification with GDL or diafiltration caused significantly lower values of total calcium and phosphate compared to the

AC

control ultrafiltered UFC (Fig. 1A and 2A). As expected, dialysis against skim milk restored the concentration of total calcium to that of the original milk for non acidified control MPC samples (UFC-D and DFC-D) (Fig. 1B). This was not the case for acidified MPC (UFG-D and DFG-D). In these samples, albeit the total calcium content increased, it remained at a lower level compared to that of skim milk (Fig. 1B).

17

ACCEPTED MANUSCRIPT After dialysis against skim milk, the total phosphate content in all MPC samples increased and reached a higher level compared to that of skim milk, in control MPC samples (UFC-D and DFC-D), with the highest concentration for DFC-D (Fig. 2B). This was not the case for the MPC prepared with GDL (UFG-D and DFG-D), where the total phosphate content of the acidified

T

sample after dialysis was still lower or similar (in the case of DFG-D) to that of skim milk (Fig.

IP

2B). The differences in the calcium and phosphate is of great importance in understanding the heat

CR

stability of the MPC dispersions.

Figures 1 and 2 summarize the concentrations of unsedimentable and permeable calcium and

US

phosphate in the MPC dispersions before or after heating. The dispersions were analyzed after

AN

heating for 10 min or at the onset of visible coagulation. There was no difference between the values measured after heating at 10 min compared to those measured in the samples at the point

M

of coagulation, for both calcium and phosphate. Therefore, the results after coagulation will not be

ED

further discussed.

All MPC dispersions before heating showed a significantly lower value of soluble (defined as

PT

the fraction that did not sediment by centrifugation) calcium (Fig. 1C) and phosphate (Fig. 2C),

CE

compared to skim milk. As expected UFG had a higher concentration of soluble calcium than UFC, as with acidification there is a higher release of calcium salts in the soluble phase (Fig. 1C and 2C,

AC

respectively). After diafiltration the amount of soluble calcium was similar amongst samples (Fig. 1C), because of the increased permeation of calcium through the membrane during diafiltration. The same trend was also observed with soluble phosphate (Fig. 2C) whereby in unheated samples, UFG had the highest soluble concentration, while UFC, DFC and DFG showed statistically similar concentrations, and lower values than UFG. This is consistent with the lower pH after heating for UFG compared to the other samples (Table 1)

18

ACCEPTED MANUSCRIPT Analysis of the same samples after dialysis against skim milk (Fig. 1D and 2D) revealed a significant effect of this process in restoring the concentrations of soluble calcium, which were all equivalent to those of skim milk (Fig. 1D). However, the concentration of soluble phosphate were lower for UFC and DFC, compared to UFG or skim milk (Fig. 2D). This may be caused by the

T

difference in the level of phosphorylated caseins present in the soluble phase.

IP

The concentration of permeable calcium and phosphate was also measured (Fig. 1 and 2, E,

CR

F). This fraction is constituted by free ions and ions linked to organic acids, and it does not contain ions associated with the soluble proteins, as proteins do not permeate through the filter. As already

US

discussed for the soluble, unsedimentable ions, in the unheated suspensions, the amounts of

AN

permeable calcium and phosphate were significantly lower in the MPC dispersions compared to skim milk. These trends were similar to what reported for the unsedimentable fraction.

M

UFG showed the highest concentration of permeable calcium and phosphate, due to the release

ED

of these ions during acidification. UFC as well as the diafiltered concentrate fractions had a low concentration of permeable ions (Fig. 1E and 2E). After dialysis against milk, the amount of

PT

permeable calcium and phosphate was restored to the concentrations of the original skim milk

CE

(Fig. 1F and 2F).

After heating (either for 10 min or at the heating coagulation time), MPC control dispersions

AC

(UFC and DFC) showed a higher amount of soluble calcium compared to their unheated counterpart (Fig. 1C). However, the amount of permeable calcium was higher only in the case of the diafiltered DFC. This suggested that the calcium present in the soluble phase was either free or associated with protein aggregates. Furthermore, there were significant differences between the amount of unsedimentable calcium after 10 min compared to that measured at heat coagulation time, for UFC (Fig. 1C): at the coagulation time, precipitation of calcium occurred. The acidified

19

ACCEPTED MANUSCRIPT concentrates did not show a higher amount of calcium after heating, but in all cases, comparable amounts to those of the original suspension (Fig. 1C and 1E). There was also a higher amount of unsedimentable and permeable phosphate (Fig. 2C) after heating of UFC. In this case, there were no differences between the soluble and permeable

T

phosphate concentrations after heating for 10 min and at the heat coagulation time. Once again,

IP

these results indicated the release of phosphate during heating in these suspensions.

CR

After dialysis against milk (Fig. 1D and 2D), as expected, there were no statistically significant differences observed in the concentrations of soluble and permeable calcium and phosphate of all

US

the MPC samples compared to skim milk. Only in the case of UFC-D (Figure 1E) there was a

AN

higher amount of soluble phosphate after heating compared to before. Similar trends were also noted for permeable calcium and phosphate (Fig. 1E and 2E) in UFC-D, although the values were

M

lower than those of soluble ions. These results suggest the presence of phosphate precipitation with

ED

heating in reconstituted UFC.

The amount of colloidal calcium and phosphate present in the reconstituted suspension was

PT

derived by difference between total and soluble salts and adjusted for the total protein

CE

concentration. The results, before and after heating at 120°C for 10 min are summarized in Table 4. In general, there was no difference in the colloidal calcium concentration after heating at 120

AC

°C for 10 min or at the onset of visible coagulation, hence, only the data after 10 min are shown in Table 4. The ratio of colloidal calcium per total protein in the reconstituted acidified MPC samples (UFG and DFG) was significantly lower compared to that of SM, as a result of the significant release of colloidal calcium from the casein micelles at a pH lower than that of native milk. Control samples (UFC and DFC) showed an intermediate value of colloidal calcium, indicating that also

20

ACCEPTED MANUSCRIPT in this case, reconstitution to a protein concentration close to that of the original milk also caused calcium dissociation from the micelles. Dialysis increased the colloidal calcium levels, but did not restore them to the original ratios in the case of UFG-D and DFG-D. In these suspensions, the ratio of colloidal calcium per total

T

protein was still significantly lower compared to that of SM. Dialysis in control MPC (UFC-D

IP

and DFC-D) restored the original values of colloidal calcium. Heating did not affect the colloidal

CR

calcium values in all reconstituted samples. This was also the case in the dialyzed samples. The ratio of colloidal phosphate (as calculated by difference between total and soluble

US

phosphate) per total protein of all MPC dispersions before and after dialysis is also compared to

AN

skim milk, before and after heating at 120 °C for 10 min (Table 4). Before heating, values of colloidal phosphate were lower than those of SM in all control MPC, while they were comparable

M

to SM in the pre-acidified MPC (UFG, DFG). This may suggest that calcium release caused a

ED

higher retention of phosphorylated caseins in the micelles. Dialysis increased significantly the ratio of colloidal phosphate per protein in reconstituted

PT

UFC-D and DFC-D, to values higher than those of the original milk. This result would indicate

CE

that dialysis caused a precipitation of phosphate in the colloidal phase of the reconstituted MPC control. It is important to note that dialysis also caused an increase in colloidal calcium.

The

AC

values of colloidal phosphate did not change after dialysis of reconstituted acidified MPC (UFGD and DFG-D). These values remained similar to the levels present in the original skim milk. After heating at 120 °C for 10 min, SM showed a significant increase in phosphate associated with the colloidal phase, consistent with phosphate precipitation. This was also the case for the DFC but not for the UFC. On the other hand, reconstituted UFG and DFG showed a significant decrease of colloidal phosphate after heating, probably because of the lower pH compared to

21

ACCEPTED MANUSCRIPT control MPC. Heating of dialyzed samples did not show further changes in the ratio of colloidal phosphate per protein, in spite of the increase during dialysis. 3.6 Protein content of soluble fraction To further understand the effect of heating on the reconstituted concentrates, the amount of

T

soluble (defined as unsedimentable) protein was also analyzed, before and after dialysis as well as

IP

after heating. Table 5 summarizes the concentration of total protein as well as the non-

CR

sedimentable protein recovered in the centrifugal supernatants (both before and after heating) of the reconstituted MPC suspensions compared to skim milk. All MPC dispersions had a statistically

US

equivalent total protein content compared to that of skim milk. Analysis of the centrifugal

AN

supernatants of the samples revealed that the amount of non-sedimentable protein for skim milk was lower compared to that of the MPC dispersions, indicating a significant impact of the MPC

M

manufacturing process and reconstitution on the partial dissociation of the casein micelles. Of the

ED

MPC dispersions, the partially acidified samples (both UFG and DFG) showed significantly higher concentrations of non-sedimentable proteins compared to the controls (UFC and DFC). This can

PT

be attributed to a significant influence of the acidification in further dissociation of casein micelles.

CE

Moreover, such impact was more profound in the same suspensions after dialysis against skim milk; the acidified samples after dialysis (UFG-D and DFG-D) showed the highest level of non-

AC

sedimentable protein, suggesting a significant impact of dialysis in dissociation of casein micelles which can be explained by the increase in ionic strength of the soluble phase. Regardless of the effect of diafiltration and/ or dialysis, no significant differences were observed in the amount of non-sedimentable proteins across the control samples: UFC, UFC-D, DFC, and DFC-D all exhibited statistically equivalent amounts of non-sedimentable proteins.

22

ACCEPTED MANUSCRIPT After heating at 120 ̊ C for 10 min, all MPC dispersions still showed significantly higher amounts of non-sedimentable protein compared to that of the SM heated under the same conditions. Of the heated MPC dispersions, the acidified UF and DF samples after dialysis (UFGD and DFG-D) showed higher amounts of non-sedimentable protein compared to the same samples

T

before dialysis (UFG and DFG). Moreover, the amount of non-sedimentable protein in the DFG-

IP

D samples was significantly higher compared to all other samples, except for the UFG-D. On the

CR

other hand, the heated UFG showed the lowest amount of the non-sedimentable protein of all MPC samples. DFC and DFG samples had a statistically equivalent amount of the non-sedimentable

US

protein, which was higher than that of UFG but lower than those of all other MPC samples. UFC-

AN

D and DFC-D exhibited a statistically equivalent amount of non-sedimentable protein. Table 5 also summarizes before and after heat treatment comparisons of the non-sedimentable

M

protein content for each sample: after heating SM, UFG, and DFG showed a significant decrease,

ED

while UFC, UFC-D, DFC-D exhibited a significant increase. The rest of the samples (UFG-D, DFG-D, and DFC) showed no significant differences.

PT

The decrease in non sedimentable protein observed in skim milk can be attributed to the heat

CE

denaturation of whey proteins and their cross-linking with the caseins forming sedimentable protein aggregates. For UFG and DFG samples, the decrease in non-sedimentable protein fraction

AC

was due to heat coagulation of the proteins, consistent with a very low HCT (Table 1). The significant increase in non-sedimentable protein content of the dialyzed control samples (both UFC-D and DFC-D) was due to a combined effect of heat and dialysis, inducing a higher ionic strength in the serum, hence promoting partial disruption of the casein micelles and an increase in the concentration of non-sedimentable caseins and of whey proteins casein aggregates. The significant increase in the non-sedimentable protein content of UFC observed upon heating, may

23

ACCEPTED MANUSCRIPT be explained by the high pH (6.97) of the samples. However, it is important to note that the DFC sample with a statistically equivalent pH as UFC, did not show a change in the non-sedimentable proteins, and it showed a HCT of approximately 8 min, which was shorter than the heating time of this experiment (10 min).

T

The consistency observed in the amount of non-sedimentable proteins of UFG-D and DFG-D

IP

can be related to the higher level of these proteins before heating, and may indicate a combined

CR

heat stability promoting effect of non-sedimentable proteins, the presence of calcium induced whey protein casein aggregates, and a pH near neutral. The dissociated caseins may in the presence

US

of the higher mineral content form heat stable aggregates with the whey proteins.

AN

3.7 SDS-PAGE of soluble fraction

To better understand possible changes in the protein composition in the continuous phase of

M

the dispersions, the centrifugal supernatants were analyzed by reducing SDS-PAGE, both before

ED

and after heating at 120°C for 10 min or for samples collected at onset of heat coagulation. Fig. 3 depicts the polypeptide composition of the soluble fractions for all the treatments before heating

PT

and Fig. 4 shows the protein profile of the supernatants of the samples heated at 120°C for 10

CE

min or at the onset of visible coagulation. Analysis of the unheated samples demonstrated a significantly higher concentration of soluble

AC

caseins in the supernatant of reconstituted MPC samples (Fig. 3, lanes 2 to 9) compared to skim milk (Fig. 3, lane 1), despite the equal initial protein content of 3.2 % (w/w). These results are in agreement with previous findings by Ferrer et al. (2011) as well as Sandra and Corredig (2013) who demonstrated that after membrane concentration there is an increase in soluble caseins in both fresh or reconstituted retentates.

24

ACCEPTED MANUSCRIPT Furthermore, after dialysis, an increase in the amount of casein proteins was observed in the centrifugal supernatant of the acidified MPC (Fig. 3: lanes 4 and 5 and lanes 8 and 9), suggesting that restoration of the serum mineral composition in the reconstituted MPC caused further solubilization of the proteins, especially for the acidified samples. It is important to note that these

T

acidified MPC dispersions had similar pH values before or after dialysis and the main differences

IP

among them were detected in the amount of phosphate and calcium of their serum phase (see Fig.

CR

1 and 2). After dialysis, the values of total calcium and phosphate were comparable to those present in skim milk.

US

The compositional analysis of the centrifugal supernatants of the samples after heating for 10

AN

min (Fig. 4) showed that, as observed in the unheated samples, all MPC dispersions had a significantly higher concentration of soluble caseins compared to skim milk (Fig. 4, lanes: 1 to 8

M

compared to lane 9). Moreover, very little casein was present in the supernatant of skim milk,

ED

while some amount of whey proteins was still recovered in the soluble aggregates. On the other hand, there were profound differences in the serum protein composition of the heated MPC

PT

dispersions before and after dialysis. The processing history and the ionic composition of the serum

CE

phase strongly affected the protein recovery in the centrifugal supernatants of the heated samples. UFC did not show any differences in composition of the soluble aggregates after 10 min of heating

AC

at 120 °C, when dialyzed before heating (Fig. 4, lanes 1 and 2). On the other hand, after diafiltration, DFC showed less protein in the heated samples, and a significant difference in the amount of soluble aggregates after dialysis against milk (Fig. 4, lanes 3 and 4). DFC had the highest pH value before dialysis, and a low amount of total calcium and phosphate compared to control. After dialysis, the pH value was restored. Similarly to DFC dispersions, the GDL treated MPC samples (UFG and DFG, Fig. 3, lanes 5-8) had less whey proteins present in the serum phase after

25

ACCEPTED MANUSCRIPT heating, with mostly caseins recovered in the soluble fraction of the heated non-dialyzed samples. On the other hand, the samples heated after re-equilibration by dialysis showed electrophoretic patterns that looked very similar to those of UFC. In this case, the GDL samples had lower pH values compared to skim milk, and more importantly, low values of total calcium and phosphate.

T

These results need to be discussed in light of the data reported on calcium and phosphate

IP

solubilization. The values of pH did not seem as critical as the concentration of soluble calcium

CR

and phosphate in modulating the amount and type of soluble aggregates present in the dispersions. The difference in the composition of the supernatants was related to the high dissociation of

US

calcium during pre-acidification in the case of UFG and DFG. These samples, showed an increase

AN

in soluble calcium and soluble protein before heating, but not a significant change in colloidal calcium phosphate. Heating did not show further changes in the calcium concentration, unlike

M

phosphate, which showed levels similar to SM (see Table 4). It could be hypothesized that the

ED

reconstituted MPC had different types of aggregates with the whey proteins, depending on the calcium sensitivity of the caseins present in the unsedimentable phase. Holt (1997) demonstrated

PT

that dissociated (non-sedimentable) caseins are more sensitive to calcium-induced protein

CE

aggregation compared to intact casein micelles, since the caseins mainly found inside the micelles (αs and β-caseins) are more sensitive to aggregation with calcium ions than the caseins normally

AC

located on the surface of the micelles (-casein). However, the type of soluble aggregates are different compared to fresh or reconstituted skim milk. The results shown in this work put into question the contribution of calcium and phosphate and pH to the heat stability of reconstituted concentrates. It has been previously reported (Fox, 1981) that calcium and phosphate deposition on the casein micelles is one of the major contributory causes of heat-induced precipitation of caseins. There were profound differences in the

26

ACCEPTED MANUSCRIPT precipitation of calcium and phosphate before and after heating, after dialysis, depending on the type of reconstituted MPC. In the case of pre-acidification during concentration, suspensions showed a higher soluble calcium, and a higher amount of soluble protein, which was prone to thermal aggregation at the acidic pH. However, the stability of the aggregates changed after

T

dialysis, when the pH increased, and a higher concentration of phosphate was recovered in the

IP

colloidal fraction. Furthermore, there was no significant deposition of phosphate on the casein

CR

micelles in the MPC dispersions after heating at 120 ̊ C for 10 min or at the onset of visible

US

coagulation.

AN

4. Conclusions

This work demonstrated that regardless of the effects of acidification and/ or diafiltration, all

M

MPC powders reconstituted at 3.2% (w/w) had a lower HCT compared to skim milk, indicating

ED

that the manufacturing process of the MPC and the compositional differences, such as pH, minerals

PT

and lactose, in general, impart a destabilizing effect on the thermal stability of the resulting powders after reconstitution. This study however, provided details of how changes in pH and

CE

serum composition may modulate the colloidal stability of casein micelles during heating of reconstituted MPC. Different MPC with different processing history and pH result in very

AC

different behaviour after reconstitution. The MPC powders made by concentrating partially acidified milk had a lower pH and exhibited very poor thermal stability compared to the MPC made with milk at normal pH. However, it was shown that the thermal stability of these powders not only recovered by restoration of pH and the serum composition through dialysis but it also surpassed the thermal stability of the control MPC dispersions after re-equilibration against skim milk. Ion concentration and pH are critical factors in determining heat stability, together with the

27

ACCEPTED MANUSCRIPT integrity of the casein micelles and the ion – protein aggregates interactions occurring in the soluble phase. These latter factors are much less characterized and understood in reconstituted MPC.

Acknowledgments

US

CR

IP

T

This project was partly funded by the Ontario Dairy Council (Ottawa, ON, Canada) and the Natural Science and Engineering Council of Canada (Ottawa, ON, Canada). Support was also provided by California Dairy Research Foundation (Davis, CA) and Dairy Research Incorporated (Rosemont, IL). Special thanks to Sean Vink (Dairy Products Tech. Center, California Polytechnic State University, San Luis Obispo) for assistance in the powder manufacture.

References

AN

Augustin, M.A. & Clarke, P.T. (1990). Effects of added salts on the heat stability of recombined concentrated milk. Journal of Dairy Research. 57: p. 213-226.

M

Anema, S.G. & Li, Y. (2003). Association of denatured whey proteins with casein micelles in heated reconstituted skim milk and its effect on casein micelle size. Journal of Dairy Research 70, 73-83.

PT

ED

Anema, S.G. (2007). Role of kappa-casein in the association of denatured whey proteins with casein micelles in heated reconstituted skim milk. Journal of agricultural and food chemistry 55(9), pp.3635-42.

CE

Anema, S.G. (2009). Effect of milk solids concentration on the pH, soluble calcium and soluble phosphate levels of milk during heating. Dairy Science and Technolology 89, 501-510. Crowley, S.V., Megemont, M., Gazi, I., Kelly, A.L., Huppertz, T., & O’Mahony, J.A. (2014). Heat stability of reconstituted milk protein concentrate powders. Int. Dairy J. 37, 104–110.

AC

Dalgleish, D.G. (1989). The behaviour of minerals in heated milks. Bulletin 238, International Dairy Federation, Brussels, pp. 31–34. de Kort, E., Minor, M., Snoeren, T., van Hooijdonk, T., & van der Linden, E. (2012). Effect of calcium chelators on heat coagulation and heat-induced changes of concentrated micellar casein solutions: The role of calcium-ion activity and micellar integrity. International Dairy Journal 26 (2) 112. Donato, L., Alexander, M., & Dalgleish, D.G. (2007). Acid gelation in heated and unheated milks: Interactions between serum protein complexes and the surfaces of casein micelles. Journal of Agricultural and Food Chemistry 55, 4160-4168.

28

ACCEPTED MANUSCRIPT Donato , L., Guyomarc'H , F., Amiot , S. & Dalgleish, D.G. (2007). Formation of whey protein/κcasein complexes in heated milk: Preferential reaction of whey protein with κ-casein in the casein micelles. International Dairy Journal, 17.1161-1167. Eshpari, H., Tong, P.S., & Corredig, M. (2014) Changes in the physical properties, solubility, and heat stability of milk protein concentrates prepared from partially acidified milk. Journal of Dairy Science, 97, 73994-7401

T

Fox, P.F. (1981). Heat-induced changes in milk preceding coagulation. Journal of Dairy Science, 64, 2127-2137.

IP

Fox, P.F. & Mulvihill D.M. (1982). Milk proteins: molecular, colloidal and functional properties'. Journal of Dairy Research, 49: 679-693.

CR

Griffin, M.C.A., Lyster, R.L.J., & Price, J.C. (1988). The disaggregation of calcium-depleted casein micelles. European Journal of Biochemistry. 174: p. 339-343.

US

Guyomarc'h, F., Law, A.J.R., & Dalgleish, D.G. (2003). Formation of soluble and micelle bound protein aggregates in heated milk. Journal of Agricultural and Food Chemistry 51, 4652-4660.

AN

Guyomarc'h, F., Queguiner, C., Law, A.J.R., Horne, D.S., & Dalgleish, D.G. (2003). Role of the soluble and micelle-bound heat-induced protein aggregates on network formation in acid skim milk gels. J. Agric. Food Chem. 51: 7743–7750.

M

Holt, C. (1997). The milk salts and their interaction with casein. Advanced Dairy Chemistry, 3, 233-256.

ED

McKenzie, H. A. (1971) Milk proteins: chemistry and molecular biology. Vol. 2. Academic Press, New York.

PT

Leviton, A., & Pallansch, M.J. (1962). High-temperature-short time-sterilized evaporated milk. IV. The retardation of gelation with condensed phosphates, manganous ions, polyhydric compounds, and phosphatides. Journal of Dairy Science. 45: p. 1045-1056.

CE

Li, Y. and Corredig M. (2014). Calcium release from milk concentrated by ultrafiltration and diafiltration. J. Dairy Sci. 97(9): 5294–5302.

AC

Li, Y., Dalgleish, D., & Corredig, M. (2015). Influence of heating treatment and membrane concentration on the formation of soluble aggregates. Food Research International. 3:309316. Marella, C., Salunke, P., Biswas, A.C., Kommineni, A., & L.E. Metzger. (2015). Manufacture of modified milk protein concentrate utilizing injection of carbon dioxide. Journal of Dairy Science, 98(6): 3577-3589. Metwalli, A.A.M., & van Boekel, M.A.J.S. (1996). Effect of urea on heat coagulation of milk. Netherlands Milk and Dairy Journal, 50, 459-476. Pappas, C.P. & Rothwell, J. (1991). The effects of heating, alone or in the presence of calcium or lactose, on calcium binding to milk proteins, Food Chem. 42.183–201

29

ACCEPTED MANUSCRIPT Rahimi-Yazdi, S, Ferrer, MA., & Corredig, M. (2010). Nonsuppressed ion chromatographic determination of total calcium in milk. J. Dairy Sci. 93:1788-1793 Sandra, S., Cooper, C., Alexander, M., & Corredig, M. (2011). Coagulation properties of ultrafiltered milk retentates measured using rheology and diffusing wave spectroscopy. Food Research International. 44, 951-956.

T

Sandra, S. & Corredig, M. (2013). Rennet induced gelation of reconstituted milk protein concentrates: The role of calcium and soluble proteins during reconstitution, International Dairy Journal, 29: 68-74

IP

Sauer, A. & Moraru, C.I. (2012). Heat stability of micellar casein concentrates as affected by temperature and pH. J. Dairy Sci. 95(11): 6339-6350.

CR

Singh, H., & Creamer, L.K. (1991). Aggregation and dissociation of milk protein complexes in heated reconstituted concentrated skim milks, J. Food Sci. 56 (1):238–246.

US

Singh, H. (2004). Heat stability of milk. International Journal of Dairy Technology, 57, 111-119.

AN

Singh, H. (2007). Interactions of milk proteins during the manufacture of milk powders. Lait. 87:413-23. Swaisgood, H.E. (1982). Chemistry of milk protein. In Developments in Dairy Chemistry -1, ed. P. F. Fox, Elsevier Applied Science Publishers, London, pp, 1-59.

ED

M

Uluko, H., Liu, L., Lv J.P., & Zhang, SW. (2016). Functional characteristics of milk protein concentrates and their modification. Critical Reviews in Food Science and Nutrition. 56(7) 1193-1208.

AC

CE

PT

Vasbinder, A.J. & de Kruif, C.G. (2003). Casein–whey protein interactions in heated milk: the influence of pH. International Dairy Journal. 13(8), pp.669-677.

30

ACCEPTED MANUSCRIPT Table 1. Values for heat coagulation time (HCT) at 120 °C and pH for pasteurized skim milk (SM) and MPC samples reconstituted in water at 3.2% equal protein (w/w), before and after dialysis (-D): UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). Values of pH reported before and after heating for 10 min. Values are the average and standard deviation of at least two measurements. Within a column, different superscript letters indicate statistical difference for P < 0.05.

pH HCT (min)

unheated

10 min

SM

58 ± 1.00 a

6.66 ± 0.00b

6.53 ± 0.00b

UFC

10.15 ± 0.15c

6.97 ± 0.04a

UFG

0.35 ± 0.10e

6.50 ± 0.00c

DFC

8.25 ± 0.25d

7.10 ± 0.02a

DFG

0.00 ± 0.00e

6.68 ± 0.04b

6.60 ± 0.02b

58 ± 1.00a

6.66 ± 0.00b

6.53 ± 0.00b

UFG-D

15.35 ± 0.20b

DFC-D

9.20 ± 0.20d

DFG-D

12.30 ± 0.15c

IP CR

6.95 ± 0.04a

US

AN

11.25 ± 0.10c

6.26 ± 0.01c 6.98 ± 0.04a

6.65 ± 0.01b

6.58 ± 0.00b

6.64 ± 0.00b

6.55 ± 0.01b

6.65 ± 0.01b

6.57 ± 0.00b

6.65 ± 0.00b

6.54 ± 0.00b

AC

CE

PT

ED

UFC-D

M

SM

T

Sample

Eshpari et al., 2017

31

ACCEPTED MANUSCRIPT Table 2. Apparent diameter of the casein micelles for pasteurized skim milk (SM) and MPC samples reconstituted in water at 3.2% protein (w/w), before or after dialysis (-D): UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). Values of pH reported before and after heating for 10 min. Values are the average and standard deviation of at least two measurements. Within a column, different superscript letters indicate statistical difference for P < 0.05. Apparent Diameter (nm) Sample

10 min

SM

169 ±1c

212 ± 1b

UFC

169±1c

160±1c

UFG

197±1a

N/A*

DFC

177±2b

DFG

207±9a

SM

169±1c

UFC-D

174±1b

UFG-D

202±1a

221± 5a

DFC-D

179±2b

218±3a

DFG-D

213±8a

218±1a

CR

IP

T

Unheated

167±1c

US

N/A*

212±1b

AC

CE

PT

ED

M

AN

223±1a

Eshpari et al., 2017 32

ACCEPTED MANUSCRIPT Table 3. Values for viscosity and turbidity (1/l*) of pasteurized skim milk (SM) and MPC samples reconstituted in water at 3.2% equal protein (w/w) before or after dialysis (-D): UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). Values of pH reported before and after heating for 10 min. Values are the average and standard deviation of at least two measurements. Within a column, different superscript letters indicate statistical difference for P < 0.05. Capital letters indicate differences before and after heating, within the same row. Turbidity: 1/l* (mm-1)

Viscosity (mPa.s)

No heat1

After 10 min heat2

No heat

After 10 min heat

SM

1.2±0.1a B

1.8± 0.1c A

2.2 ±0.1a A

1.8± 0.1b B

UFC

1.1±0.1a A

1.5±0.1c,d A

1.6±0.1b A

UFG

1.2±0.1a B

3.1±0.1a A

1.6±0.1b A

DFC

1.1± 0.1a A

1.3± 0.1d A

1.4±0.1b A

1.5±0.2b,c A

DFG

1.2± 0.1a B

2.5±0.1b A

1.6±0.1 b A

1.4±0.1c A

SM

1.2±0.1a B

1.8±0.1c A

2.2±0.1a A

1.8 ± 0.1b B

UFC-D

1.1±0.1a B

2.4±0.1b A

2.2±0.1a A

2.2±0.1a A

UFG-D

1.0±0.1a B

1.6±0.0c A

2.3±0.1a A

2.4±0.1a A

DFC-D

1.4±0.1a B

2.3±0.1b A

2.1±0.1a A

2.4±0.1a A

DFG-D

1.2± 0.1a B

1.7±0.1c A

2.4±0.3a A

2.6±0.1a A

AC

CE

PT

ED

M

AN

US

CR

IP

T

Sample

Eshpari et al., 2017 33

1.4± 0.1c B

1.3±0.04c A

ACCEPTED MANUSCRIPT Table 4. Values for colloidal calcium and colloidal phosphate (mg kg-1) / total protein (mg kg-1) before and after 10 min heating at 120 °C for pasteurized skim milk (SM) and MPC samples reconstituted in water at 3.2% equal protein (w/w), before and after dialysis (-D): UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). Values are the average and standard deviation of at least two measurements. Within a column, different superscript letters indicate statistical difference for P < 0.05. Pairwise comparison within a row are also identified with different superscript letters at P<0.05.

Colloidal Calcium/ Total protein heated

unheated

heated

SM

0.023 ± 0.001b A

0.025 ± 0.001b A

0.036 ± 0.001b A

0.053 ± 0.001a B

UFC

0.019 ± 0.002c A

0.018 ± 0.001c A

0.029 ± 0001c A

0.030 ± 0.002d A

UFG

0.013 ± 0.001d A

0.014 ± 0.001d A

0.036 ± 0.002b A

0.027 ± 0.001d B

DFC

0.017 ± 0.001 c A

0.017 ± 0.001c A

0.028 ± 0.002c A

0.035 ± 0.001c B

DFG

0.012 ± 0.001d A

0.013 ± 0.002d A

0.036 ± 0.001b A

0.029 ± 0.001d B

UFC-D

0.023 ± 0.001b A

0.024 ± 0.001b A

0.050 ± 0.003 a A

0.044 ± 0.002b A

UFG-D

0.015 ± 0.002c A

0.017 ± 0.001c A

0.032 ± 0.001c A

0.033 ± 0.001c A

DFC-D

0.028 ± 0.002 a A

0.029 ± 0.001a A

0.055 ± 0.002a A

0.051 ± 0.002a A

DFG-D

0.016 ± 0.001c A

0.018 ± 0.001c A

0.038 ± 0.001b A

0.036 ± 0.002c A

AC

CE

PT

ED

M

US

CR

IP

T

unheated

AN

Sample

Colloidal phosphate/ Total protein

Eshpari et al., 2017

34

ACCEPTED MANUSCRIPT Table 5. Values (%) of total and non-sedimentable protein concentrations (both before and after heating at 120 °C for 10 min) for pasteurized skim milk (SM) and MPC samples reconstituted in water at 3.2% equal protein (w/w): UF control before dialysis = UFC; acidified UF before dialysis = UFG; DF control before dialysis = DFC; acidified DF before dialysis = DFG; UF control after dialysis = UFC-D; acidified UF after dialysis = UFG-D; DF control after dialysis = DFC-D; acidified DF after dialysis = DFG-D. Values are the average and standard deviation of at least two measurements. Within a column, different superscript letters indicate statistical difference for P < 0.05. Capital letters indicate differences in the non-sedimentable protein fraction of the samples before and after heating, within the same row. Total protein (%)

Supernatant protein (non-heated) (%)

Supernatant protein (heated) (%)

SM

3.21 ± 0.03 a

0.84 ± 0.02 d A

0.35 ± 0.03 f B

UFC

3.20 ± 0.08 a

1.75 ± 0.05 c B

UFG

3.20 ± 0.01 a

2.54 ± 0.02 b A

DFC

3.21 ± 0.01 a

1.79 ± 0.02 c A

DFG

3.20 ± 0.03 a

2.62 ± 0.03 b A

1.59 ± 0.02 d B

UFC-D

3.16 ± 0.02 a

1.80 ± 0.03 c B

2.78 ± 0.01 bc A

UFG-D

3.18 ± 0.03 a

2.94 ± 0.01 a A

2.92 ± 0.02 ab A

DFC-D

3.20 ± 0.02 a

1.86 ± 0.01 c B

2.80 ± 0.01 bc A

DFG-D

3.19 ± 0.01 a

2.95 ± 0.02 a A

2.96 ± 0.01 a A

2.71 ± 0.04 c A 1.05 ± 0.03 e B 1.67 ± 0.02 d A

Supernatant protein (non-heated)

Supernatant protein (heated 10 min)

(%)

(%)

(%)

0.84 ± 0.02 d A

0.35 ± 0.03 f B

3.20 ± 0.08 a

1.75 ± 0.05 c B

2.71 ± 0.04 c A

3.16 ± 0.02 a

1.80 ± 0.03 c B

2.78 ± 0.01 bc A

UFG

3.20 ± 0.01 a

2.54 ± 0.02 b A

1.05 ± 0.03 e B

UFG-D

3.18 ± 0.03 a

2.94 ± 0.01 a A

2.92 ± 0.02 ab A

DFC

3.21 ± 0.01 a

1.79 ± 0.02 c A

1.67 ± 0.02 d A

DFC-D

3.20 ± 0.02 a

1.86 ± 0.01 c B

2.80 ± 0.01 bc A

DFG 3.20 ± 0.03 a Eshpari et al., 2017

2.62 ± 0.03 b A

1.59 ± 0.02 d B

3.19 ± 0.01 a

2.95 ± 0.02 a A

2.96 ± 0.01 a A

SM UFC UFC-D

DFG-D

CE

Total protein

3.21 ± 0.03 a

AC

Sample

PT

ED

M

AN

US

CR

IP

T

Sample

35

ACCEPTED MANUSCRIPT

1400

1000

b

bc

800

bc

c

600 400 200

1200 800 600 400 200 SM

Soluble calcium (mg/kg)

UFC

c

UFG

bc

b

c

DFC

300 250

c cc

50 0 SM

PT

b bb

200 100

DFG

E

aa

150

bc

M

a

c

UFC

ED

400

bb

AN

c

c d

UFG

d

DFC

c

cd

cd

Permeable calcium (mg/kg)

b b

a

aa

a b

a

a

ba

a

ba

a a b

D

US

aa

450 400 350 300 250 200 150 100 50 0

UFC -D UFG -D DFC -D DFG -D

IP

DFG

CR

DFC

C

CE

Soluble calcium (mg/kg) Permeable calcium (mg/kg)

UFG

a

SM

350

UFC

b

b

0 SM

B

a

a

1000

0

450 400 350 300 250 200 150 100 50 0

a

T

A

a

1200

Total calcium (mg/kg)

Total calcium (mg/kg)

1400

400 350 300 250 200 150 100 50 0

SM

a

a a

SM

DFG

UFC -D UFG -D

DFC -D

DFG -D

F

aa

a aaa

a aa

aaa

UFC -D UFG -D DFC -D DFG -D

AC

Fig. 1. Total, soluble, and permeable calcium for skim milk (SM) and various MPC resuspended to 3.2% (w/w) protein, before (A, C, and E) or after (B, D, and F) dialysis (-D). UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). White bar: unheated, Black bar: heated at 120°C for 10 min, and Patterned bar: onset of visible coagulation at120°C. Different letters indicate significant differences in each calcium fraction across different samples . Error bars represent standard deviation

Eshpari et al., 2017 36

ACCEPTED MANUSCRIPT

b

b

c

1000 500 0 UFG

DFC

C Soluble phosphate (mg/kg)

a

c

600

b

b bb

b

400 200

ccc ccc

0

1000

UFC

UFG

a

DFG

E

800 600

DFC

M

SM

a a

400

c

b b bbb

0

UFC

UFG

c c d c d c

DFC

DFG

0

1600 1400 1200 1000

a

UFC -D UFG -D DFC -D DFG -D

c

D b bb b b abb b ab bc

aa

800 600 400 200 0

SM

UFC -D UFG -D DFC -D DFG -D

1200 1000 800 600

a

a

bb

a

bb

a

a bb

F

bb

aa

400 200 0 SM

AC

SM

CE

200

500

AN

800

1000

US

aa

1000

1500

SM

1400 1200

2000

DFG

Permeable phosphate (mg/kg)

1600

UFC

cd

d

2500

T

b

1500

B

a

b

c

3000

IP

2000

ED

Soluble phosphate (mg/kg)

Total phosphate (mg/kg)

2500

SM

Permeable phosphte (mg/kg)

3500

A

CR

a

PT

Total phosphate (mg/kg)

3000

UFC -D UFG -D DFC -D DFG -D

Fig. 2. Total, soluble, and permeable phosphate for skim milk (SM) and various MPC resuspended to 3.2% (w/w) protein, before (A, C, and E) or after (B, D, and F) dialysis (-D). UF control (UFC) or acidified (UFG); DF control (DFC) or acidified (DFG). White bar: unheated, Black bar: heated at 120°C for 10 min, and patterned bar: onset of visible coagulation at120°C. Different letters indicate significant differences in each calcium fraction across different samples . Error bars represent standard deviations and permeable

Eshpari et al., 2017 37

ACCEPTED MANUSCRIPT

>mW Peptides αs CN β CN β-lg

2

3

4

5

6

7

8

9

CR

1

IP

T

α-la

AC

CE

PT

ED

M

AN

US

Fig. 3. SDS-PAGE of centrifugal supernatants (10 μl) of pasteurized skim milk (lane 1) and reconstituted MPC (lane 2-9). All samples were reconstituted at 3.2% (w/w) protein. UF control: UFC (lane 2), UFC-D (lane 3); UFacidified: UFG (lane 4), UFG-D (lane 5); DF control: DFC (lane 6), DFC-D (lane 7); DF acidified: DFG (lane 8), DFG-D (lane 9). Results are representative of three replicate runs.

Eshpari et al., 2017

38

T

ACCEPTED MANUSCRIPT

CR

IP

>mW Peptides αs CN

2

3

4

5

6

7

AN

* 1

US

β CN

8

β-lg α-la

9

PT

ED

M

Fig. 4. SDS-PAGE of centrifugal supernatants (10 μl) of pasteurized skim milk (lane 9) and reconstituted MPC(lane 1-8). All samples were reconstituted at 3.2% (w/w) protein and heated at 120 °C for 10 min. UF control: UFC (lane 1), UFC-D (lane 2); UFacidified: UFG (lane 3), UFG-D (lane 4); DF control: DFC (lane 5), DFC-D (lane 6); DF acidified: DFG (lane 7), DFG-D (lane 8). Low molecular weight standard (lane *). Results are representative of three replicate runs.

AC

CE

Figure 6. 5 SDS-PAGE of centrifugal supernatants (10 μl) of pasteurized skim milk (lane 9) and reconstituted MPC(lane 1-8). All samples were reconstituted at 3.2% (w/w) protein and heated at 120 °C for 10 min. UF control: UFC (lane 1), UFC-D (lane 2); UFacidified: UFG (lane 3), UFG-D (lane 4); DF control: DFC (lane 5), DFC-D (lane 6); DF acidified: DFG (lane 7), DFG-D (lane 8). Low molecular weight standard (Lane *). Results are representative of three replicate runs.

Eshpari et al., 2017 39

ACCEPTED MANUSCRIPT Graphical abstract

Ultrafiltration Diafiltration Preacidification

   

Heating 120°C

MPC

Heat coagulation time Ionic balance Turbidity Protein composition

CE

PT

ED

M

AN

US

CR

IP

T

reconstitution

AC

  

40

ACCEPTED MANUSCRIPT Highlights 

Milk was ultrafiltered to a target pH of about 6.0, and different MPC powders obtained



All MPC powders showed less heat stability than skim milk control at the same protein concentration. Heat stability of reconstituted MPC decreased with preacidification



Heat stability of preacidified MPC improved if the original milk serum composition was

IP

T



US

Treatment of MPC at 120° C did not cause significant deposition of calcium and phosphate

CE

PT

ED

M

AN

on the micelles.

AC



CR

reestablished.

41