Colour change and proteolysis of skim milk during high pressure thermal–processing

Journal of Food Engineering 147 (2015) 102–110

Contents lists available at ScienceDirect

Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng

Colour change and proteolysis of skim milk during high pressure thermal–processing Anastasia Fitria Devi a,b, Roman Buckow b,⇑, Tanoj Singh b, Yacine Hemar c, Stefan Kasapis a a

School of Applied Sciences, RMIT University, VIC 3001, Australia CSIRO Animal, Food and Health Sciences, Werribee, VIC 3030, Australia c School of Chemical Sciences, The University of Auckland, AUC 1142, New Zealand b

a r t i c l e

i n f o

Article history: Received 13 April 2014 Received in revised form 25 August 2014 Accepted 10 September 2014 Available online 21 September 2014 Keywords: High pressure Skim milk Browning Proteolysis Sugar conjugation Kinetics

a b s t r a c t The effect of high pressure–thermal processing (HPTP), in the range of 0.1–600 MPa and 100–140 °C for up to 60 min, on colour change and proteolysis of reconstituted skim milk (10% w/w) was investigated. The kinetic results showed that colour change (DEab) and proteolysis (determined by chromatography) increased with both increasing temperature and pressure. The apparent reduction of free amino groups in skim milk, indicating sugar conjugation to milk proteins/peptides, was accelerated with increasing temperature, but decelerated with increasing pressure (at constant temperature) at higher temperature. The milk’s colour changed drastically at 400 MPa where most of the milk proteins formed coagulates and left the solutions nearly translucent. Mathematical models describing the kinetics of colour change, proteolysis, and free amino acids reduction as a function of pressure and temperature are proposed. Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

1. Introduction Untreated raw milk is highly perishable and often contains food-borne bacterial pathogens. Thus, selling of raw milk is prohibited or at least restricted to direct commercialisation at the farm in most countries. Sterilisation and long shelf life of milk is often achieved through ultra high temperature (UHT) processing for a few seconds at or above 135 °C; however, canning of milk products at 120 °C up to 30 min is still practiced in the dairy industry (Kessler, 1996). Milk experiences colour change during thermal processing. Firstly, an increase in lightness occurs due to denaturation of b-lactoglobulin and its conjugation to j-casein (Burton, 1994). At the early stage of the Maillard reaction in milk, lysine conjugates with lactose and results in colourless compounds. These compounds convert into coloured products including melanoidins during the final stage of the Maillard reaction (Burton, 1994). Compared to other heat induced changes of milk systems (e.g., Maillard reaction and disulfide-bonded aggregation), only a few studies have been carried out on heat induced proteolysis. One of the reasons is the lack of high precision analytical techniques for identifying the proteolysis products and determining their ⇑ Corresponding author. Tel.: +61 3 97313270; fax: +61 3 9731 3202. E-mail address: [email protected] (R. Buckow). http://dx.doi.org/10.1016/j.jfoodeng.2014.09.017 0260-8774/Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

concentrations (Morales and Jiménez-Pérez, 1998). Most studies on thermal proteolysis of milk proteins focus on casein since casein is more heat stable than whey protein and more available in milk (Guo et al., 1989). Casein degrades during severe heating (i.e., P120 °C) and results in peptide fractions, mostly from as1-casein followed by j-casein (Hustinx et al., 1997; Gaucheron et al., 1999). Whey proteins strengthen their hydrophobic bonding during heating to 60 °C. They unfold and denature during heating between 60 and 100 °C. Further heating to 140 °C results in breakdown of disulfide bonds and reduction of covalent cross-links (de Wit and Klarenbeek, 1984). High pressure–thermal processing (HPTP) combines high pressure (up to 800 MPa) and heating (above 60 °C) has been considered for sterilisation and shelf life extension of foodstuff due to its ability to inactivate bacterial spores at reduced heat and thereby preserving desirable functional properties of foods better than conventional thermal processing (Heinz and Buckow, 2010). Accelerated and homogeneous heating and cooling of food occurs during HPTP due to the increase and decrease in temperature accompanying the physical compression and decompression of the product. This facilitates uniform heating of all food and also reduces the need for excessively long heating times and often results in improved food quality attributes, such as flavour, texture, nutrient content, and colour, compared with conventional thermal processing (Grauwet et al., 2012).

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

Studies addressing the effect of HPTP on the quality of dairy products are still limited (Devi et al., 2013). In contrast to heated milk, the research on visual change of high pressure (applied at 640 °C) treated milk focused on its light scattering properties (i.e., lightness and turbidity) rather than its tristimulus colour variables (i.e., Lab system). Reduced lightness and turbidity of milk upon high pressure processing (HPP) is usually associated with changes of casein micelle conformation (Huppertz et al., 2002). The processing temperature (4–40 °C) during HPP, particularly at moderate pressures (200–300 MPa), affects the lightness/turbidity of milk. The differences in the optical properties of milk after HPP at 250 MPa at various temperatures (4–40 °C) might be due to temperature-dependent casein micelle dissociation (during pressurisation) or reassociation (during pressure release), or combination of both (Gaucheron et al., 1997; Orlien et al., 2006). Under HPTP conditions, a more complex interaction amongst milk constituents (e.g., proteins and lactose) is likely to take place which can result in the milk’s concomitant colour change due to Maillard reaction products (Jaeger et al., 2010). To the best of our knowledge, studies focusing on the quality of bovine milk as affected by HPTP at higher temperatures (i.e., P100 °C) are not available. Therefore, the aim of this kinetic study was to investigate and predict proteolysis in skim milk, interaction between milk proteins with lactose, and consequences to skim milk colour during HPTP and heat-only treatments at 100–140 °C.

103

tainers containing boiling milk was not possible. Skim milk was filled with no airspace into cryogenic vials. The cap of a duplicate vial was pierced with a needle to insert a thermocouple. Samples were sealed and stored at approximately 4 °C before they were placed into the pressure vessels heated to target temperatures. Pressurisation was started when the sample reached a temperature which would result in the target temperature after compression heating. The compression rate was set to 21 MPa/s to achieve an operational pressure of 600 MPa in less than 30 s. Pre-heating times were very short (<60 s) but vary slightly depending on the applied pressure–temperature conditions. Treatment time was started as soon as isobaric and isothermal conditions were reached. A data acquisition system (2700 Integra multimeter, Keithley Instruments, Cleveland, OH, USA) connected to the high pressure multi-vessel apparatus U111 software for data acquisition (Version 2.1c, Unipress, Warsaw, Poland) was used to monitor the pressure and temperature history of each sample. The initial time (0 min) was defined as the time point when target temperature and/or pressure were reached, followed by immediate decompression and/or cooling in an iced water bath. The initial colour, initial area (A0), and initial free amino group concentration (C0) were defined as the colour, area, and concentration, respectively, found in the samples at 0 min. Area (A) refers to the area under the peptide chromatograms in the HPLC spectrum and was calculated using mathematical integration. All experiments were performed at least in duplicate.

2. Materials and methods 2.4. Colour measurement 2.1. Material Skim milk powder (SMP), produced with low-heat, was purchased from Tatura Milk Industries Ltd. (Tatura, VIC, Australia). The composition according to the manufacturer was as follows: 33.6% protein, 0.7% fat, 2.8% ash, 3.6% moisture, and 59.3% lactose. A solution of 10% w/w in deionised water gave a pH value of 6.7. 2.2. Sample preparations Reconstituted skim milk at 10% w/w was used throughout the study and was prepared daily and separately for each experiment. SMP was dispersed in deionised water under stirring for 2 h at room temperature and allowed to fully hydrate overnight at 5 °C. Milk solutions were transferred into 2.0 mL crimp vials (#5181-3375, Agilent Technologies Inc., Santa Clara, CA, USA) and sealed with an 11 mm silver aluminium crimp FEP/rubber cap (#5181-1210, Agilent Technologies) prior to thermal processing. For HPTP, milk solutions were filled into 1.0 mL cryogenic vials (#5000–1012, Nalgene, Rochester, NY, USA). 2.3. Isothermal–isobaric treatments Milk solutions were treated under isothermal/isobaric conditions at temperatures ranging from 100 to 140 °C and pressures of 0.1–600 MPa up to 60 min to accurately follow and understand selected biochemical reactions in milk. Heat-only treatments were performed in a temperature-controlled shaking glycerol bath (#SWB20, Ratek, Boronia, VIC, Australia). Heating of the glass vials to target temperature took approximately 1 min. Following treatment, the samples were immediately cooled in an iced water bath and chemical analysis was performed within 4 h. HPTP experiments under isothermal conditions were performed using a multi-vessel high pressure unit (#U111, Unipress, Warsaw, Poland) as described previously (Buckow et al., 2011). The target temperature was varied between 100 and 130 °C and the pressure of 200, 400 or 600 MPa was applied. Application of 200 or 400 MPa at 130 °C was not feasible as the removal of (flexible) plastic con-

Colour of milk samples was determined at room temperature using a chromameter (Konica Minolta CR-300, Morinuchi, Tokyo, Japan) and recorded in CIE-Lab tristimulus system. Milk was placed in a glass cuvette, inserted into a black chamber (provided by Konica Minolta), and connected to the chromameter. This arrangement provided 90° angle of observation. Standard illuminant D65 was used as the light source. Colour measurement was taken in triplicate for each sample and average values were used. The colour difference between milk before and after the treatment was expressed as colour difference (DEab), which was calculated using Eq. (1).

DEab ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ½ðL2  L1 Þ2 þ ða2  a1 Þ2 þ ðb2  b1 Þ 

ð1Þ

where L is the lightness, a is the green–red co-ordinate, while b is the blue–yellow co-ordinate. 2.5. HPLC analysis Prior to separation of peptides on HPLC, all samples were adjusted to 2% w/v trichloroacetic acid (TCA), by addition of 167 lL of 8% w/v TCA (#200-927-2, AnalaR, Merck Millipore, Kilsyth, VIC 3137, Australia) to 500 lL of milk sample. Short time (3  5 s) homogenisation (IKA Labortechnik T8 Ultra Turrax Disperser, Staufen, Germany) at pulse No. 2 was applied on coagulated samples before 500 lL was taken for acid precipitation. The precipitated large peptides and intact milk proteins were removed by centrifugation (25,000  g for 20 min at 4 °C) and the TCA soluble peptides were filtered through a 0.45 lm filter (#2165, GRACE Davison Discovery Sciences, Rowville, VIC 3178, Australia) before application to the column. HPLC was performed using an automated Thermo Finnigan Surveyor Plus system (San Jose, CA, USA) fitted with a widepore C18 reversed phase Aeris column (particle size 3.6 lm, pore size 300 Å, 150 mm  2.1 mm; Phenomenex, Lane Cove, NSW 2066, Australia) and guard column (10 mm  2.1 mm, Phenomenex). The column temperature was maintained at 35 °C.

104

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

Two solvents were used for elution; solvent A was 0.1% v/v of trifluoroacetic acid (TFA, #302031, Sigma–Aldrich, Castle Hill, NSW 1765, Australia) in water and solvent B was 0.1% v/v of TFA in acetonitrile (#OS-10520, Optigen Scientific, Jacksonville, FL, USA). Peptide separation was achieved with a linear gradient of 2–70% solvent B (0.08% TFA in acetonitrile) in solvent A (0.1% TFA in water) over 65 min. The flow rate was maintained at 150 lL min1. The eluate was monitored at 214 nm using a photodiode array detector. The HPLC system was connected to a computer using Xcalibur software (v1.4, Thermo Scientific, Waltham, MA, USA). 2.6. O-phthaldialdehyde/N-acetyl-L-cysteine (OPA/NAC) assay Measurement of free amino groups of milk samples was performed using an OPA/NAC method (Hernández et al., 1990), modified for microplates and fluorescence readers (Lochmann et al., 2004). OPA (#P0657, Sigma Aldrich) and NAC (#A7250, Sigma Aldrich) solutions were prepared separately in ethanol and water, respectively, at a concentration of 0.05 M. Two solutions were prepared weekly and kept at 5 °C in the dark prior to further usage. Boric acid 0.1 M (#10058.3R, AnalaR, Merck Millipore) buffer was prepared and adjusted to pH 9.5 by adding 1 M NaOH solution. OPA/NAC reagent was prepared daily by gently mixing OPA, NAC, and boric acid buffer solutions with a volumetric ratio of 1:1:8, respectively. This reagent was stored in the dark at 5 °C for at least 2 h to eliminate the background fluorescence (Lochmann et al., 2004). Milk samples were diluted 500 fold by addition of water to be within the linear range of the assay and 100 lm was pipetted into the wells of a black Optiplate-96 F microplate (#81-635, Perkin–Elmer, Walthan, MA, USA). Homogenisation of milk samples was performed as described previously in cases where milk was clotted during treatments. A flash microplate reader (4.00.25, Thermo Fisher Scientific, Waltham, MA, USA) with SkanIt software (v2.4.3.37) was used to perform the fluorometric measurements. After a short incubation period at 25 °C, 200 lL of OPA/NAC reagent was automatically dispensed at medium speed into each well. The microplate was shaken at 360 spm for 5 min followed by 5 min settle delay for the completion of reaction. The reading was performed at an excitation wavelength of 340 nm and an emission wavelength of 434 nm, with an excitation bandwidth of 12 nm for 100 ms. The fraction of free amino groups (C/C0) was the relative fluorescence unit (RFU) of milk sample after treatment divided by their initial RFU (Morales et al., 1995).

from Baranyi et al., 1999) were estimated. The former informs how well the data fit the model whilst the latter indicates the deviation between the model predictions and the observed results. A perfect agreement of the model to the fitted data is represented by R2 equals to 1. A smaller value of Af indicates a more accurate absolute error in the model.

P 2 ½Lnðkobs Þ  Lnðkpred Þ R2 ¼ 1  Pi 2  i ½Lnðkobs Þ  Lnðkobs Þ

ð2Þ

where kpred is the predicted coefficient calculated based on the secondary model, kobs is the observed coefficient at a given pressure–  temperature combination, and k obs is the mean of the observed coefficients.

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! SE ðM  NÞ Af ¼ exp M

ð3Þ

where SE is the standard error of fit, M is the number of observations, and N is the number of fitted parameters. 3. Results and discussion 3.1. Colour change Milk experienced colour change from white to brown during isothermal heating at ambient and elevated pressure (Fig. 1). Brown colour formation is due to both Maillard and caramelisation reactions (Morales and van Boekel, 1998). Coloured Maillard products could be grouped into low molecular weight compounds (<500 Da) of which some of them have been characterised (Rizzi, 1997), and polymeric melanoidins with molecular weights up to 100,000 Da of which chemical structures are largely unknown (van Boekel, 1998). Melanoidins contain nitrogen which commonly provided by e-amino group on lysyl residues in the primary amino acid sequence of milk proteins. The free e-amino group on the side chain of lysine decreases dramatically during the final (browning) stage of Maillard reactions in milk (van Boekel, 1998). Brown pigments, however, can be formed without nitrogen; for example, in the case of sugar degradation (van Boekel, 1998). Samples treated at 400 MPa and 110 °C (Fig. 1) showed precipitation due to protein coagulation resulting in less particles being

2.7. Kinetic data analysis Kinetic models are used to describe colour change, proteolysis, and free amino groups reduction because it can describe quality changes with time. The kinetic data sets (e.g., colour change, OPA, and proteolysis) from each pressure–temperature combination were fitted to mathematical models which gave minimum cumulative sum of standard error of fit (RSE) using Table Curve 2D (v.4.07, Sysstat Software Inc., Chicago, IL, USA). Each primary model features a coefficient that changes depending on the applied pressure–temperature combination. Therefore, the obtained coefficient values were fitted (Table Curve 3D, v.3.01, Sysstat Software, Inc., Chicago, IL, USA) to a secondary model as a function of pressure and temperature. The model which gave minimum standard error of fit (SE) was then selected as the secondary model. The mathematical models should not give significant heteroskedasticity, which was displayed by small deviation of the predicted versus observed coefficients from the bisector line in the parity plot. For examining the performance of the secondary model, the coefficient of determination (R2) (Eq. (2)) and the accuracy factor (Af) (Eq. (3), taken

Fig. 1. Colour change of skim milk from white to caramel brown during HPTP at 400 MPa and 100, 110 and 120 °C, respectively. Samples with treatment time zero were pressurised to target pressure and temperature followed by immediate cooling in an iced water bath. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

present in the soluble phase. Protein coagulation was also obtained for samples treated at 400 MPa and 100 °C for more than 5 min (Fig. 1). Visible aggregates instead of coagulation were observed in the samples treated at 400 MPa and 120 °C for more than 5 min. Overall, treatments at 400 MPa and 100–120 °C resulted in low lightness and the highest colour difference of these translucent milks compared to other samples treated at the same temperatures. Protein complexes in our initial milk samples were possibly those naturally present in raw milk (Chevalier and Kelly, 2010) and those formed during spray drying (Patel et al., 2007) and storage (Le et al., 2012). Protein complexes in raw skim milk mainly consist of aS2-casein and j-casein homopolymers, aS2-casein/ j-casein heteropolymers, and other heteropolymers (Chevalier and Kelly, 2010). Spray drying facilitated the formation of not only protein complexes (i.e., j-casein/b-lactoglobulin complexes) but also protein–sugar conjugates (i.e., j-casein and lactose) and other Maillard products (Le et al., 2012). Severe heating (i.e., P100 °C) activates the nucleophilic amino acid residues such as lysine and cysteine of milk proteins to form intra and intermolecular bonds. Complexes may be formed through various types of bonds which was addressed by O’Connell and Fox (2003). The mechanism behind the heat induced polymerisation, however, is not yet fully elucidated. In our study, more aggregates were formed during HPTP. They could be linked through intra and intermolecular disulfide bonds especially when aS2-casein, j-casein, and whey protein with cysteine residues are involved (Patel et al., 2006; Chevalier and Kelly, 2010). Protein without cysteine residues (aS1-casein and b-casein) probably link to the complexes via non-disulfide bonds, such as glycation cross-link formed during Maillard reaction (Chevalier et al., 2009). We also propose that aS2-casein and a-lactalbumin were incorporated into protein complexes in our samples. Since aS2-casein is not a surface component of the micelle, such high temperature is required to assist dephosporisation of casein micelle (Metwalli and van Boekel, 1998), and therefore aS2-casein can form complexes with whey proteins. Complexation of aS2-casein with j-casein and/or whey proteins requires application of high temperature (i.e., 120 °C for 120 s) (Patel et al., 2006). The temperature ranges used in our study are more severe and thus, surpass the critical condition required to facilitate the complexation of aS2-casein with other proteins. Complexes of aS2-casein/j-casein/b-lactoglobulin, however, were resulted from skim milk after HPP (400 MPa for 30 min at 22 °C) (Patel et al., 2006). The temperature should not affect the aggregate profile of skim milk and should not be adequate to assist the accessibility of aS2-casein. Thus, HPP (400 MPa) induces the unfolding of b-lactoglobulin and its aggregation with other b-lactoglobulin molecules and/or j-casein. It also facilitates the dissociation of micelle, and thus aS2-casein is accessible by b-lactoglobulin. More aS2-casein is involved in the complexes during HPP P600 MPa. Meanwhile, only a small percentage (13%) of a-lactalbumin is involved in the heat induced (90 °C for 30 min) complexes (Chevalier and Kelly, 2010). HPP (600 MPa for 30 min at 22 °C) induces the incorporation of a-lactalbumin into the complexes, although to a lesser extent compared to b-lactoglobulin (Patel et al., 2006). It was believed that it was due to less exposed disulfide bonds of a-lactalbumin during and after HPP. Aggregates containing a-lactalbumin were also obtained in skim milk after HPTP (500 MPa and 55 °C for 5 min or 600 MPa and 70 °C for 5 min). aS2-Casein, however, was not detected in the aggregates (Nabhan et al., 2004). To the best of our knowledge, no published information is available on milk aggregation during HPTP at and above 100 °C. Another factor which possibly contributed to protein coagulation of our processed samples is the decrease of milk’s pH. During

105

heating, acid is produced, mostly from lactose degradation. Most (80%) of the acid is formic acid (O’Connell and Fox, 2003). The pH of milk upon heating is usually measured after a fixed cooling time and temperature because the measurement at the actual heating condition sometimes would not be possible, for example the measurement of pH of milk during heating at 140 °C, and thus extrapolation is used for prediction. Heating at 140 °C for 5 min without any cooling stage resulted in the milk’s pH decline from 6.7 to 5.5. Extended heating at 140 °C for 20 min reduced the pH of milk to 4.9 (Fox, 1981). This pH is close to the acidcoagulation point (pH  4.6) of milk. Thus, the repulsion force between milk proteins is less. HPP alone increases the pH of skim milk. According to Schrader and Buchheim (1998), the pH of UHT skim milk went up by 0.11, upon HPP (400 MPa, 10 min, room temperature). During storage at room temperature for 4 h, the pressurised milk reached its final pH, about 0.05 units higher than its initial value. This minor pH increase can be ignored considering a remarkable pH decrease due to heating. 3.2. Modelling colour change The modelling of colour change of skim milk in this study is based on the Lab values, regardless the changes in turbidity and the present aggregation/coagulation. Each kinetic curve was fitted to Eq. (4). The numerator was estimated by evaluating values from P 31 to 45 and selecting the value 38 giving the lowest SE.

DEab ¼

38t Gþt

ð4Þ

where G is the coefficient in the primary model for colour change and t (min) is the treatment time. A linear change in colour difference DEab of skim milk was observed at 100 and 110 °C up to 60 min. At higher temperature, the colour changed more rapidly and appeared linear only in the first 10 min (Fig. 2A). A study by Pagliarini et al. (1990) showed that the increase of DEab in fresh skim milk during heating at 90–120 °C for 30 min follow a zero-order reaction. The differences between their result and ours probably due to differences in the determination of initial milk colour. Our investigation revealed that under pressure, the colour change in milk from white to caramel brown was accelerated at constant temperature (Fig. 2B). The estimated values of G for all kinetic curves are shown in Table 1. The pressure–temperature dependence of G was then estimated by non-linear regression fitting of the data to a polynomial model (Eq. (5)) using p0 = 400 MPa and T0 = 110 °C (Table 2).

Ln G ¼ G0 þ G1 ðp  p0 Þ þ G2 ðT  T 0 Þ þ G3 ðp  p0 Þ2 þ G4 ðp  p0 Þ3 þ G5 ðT  T 0 Þðp  p0 Þ2

ð5Þ

where G0, G1, G2, G3, G4, and G5 are the coefficients in the secondary model of colour change while p0 and T0 are the reference pressure and temperature, respectively. The parity plot (Fig. 3A) of the natural logarithm of the experimental versus the predicted G values indicates no significant heteroskedasticity during the mathematical modelling of milk colour change as the deviations from the bisector are small. The model has a R2 of 0.966 and an Af of 1.268. High pressure can accelerate the colour development, for example, heating at 131 °C at ambient pressure for 30 min is required to reach DEab of 25, whereas at 300 MPa, 30 min treatment at 115 °C will result in the same DEab. Heating at 115 °C and ambient pressure results in a DEab value of only 10. The drawback of this pressure–temperature diagram (not shown) is it indicates that the colour change in skim milk is hampered under mild high pressures (50 MPa), where no observations were performed and, thus, this

106

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

Fig. 2. Colour change of reconstituted skim milk (10% w/w) during thermal treatment at (A) 100–140 °C and 0.1 MPa and (B) during HPTP at 110 °C and 0.1–600 MPa as expressed by total colour difference (DEab). Solid lines are obtained by fitting the data into Eq. (4).

Table 1 Estimated values of coefficient G (102) in Eq. (4) for combined high pressure (p in MPa) and heat (T in °C) induced colour change in reconstituted skim milk (10% w/w).

a b

p/T

100

110

120

130

140

0.1 200 400 600

3.708 ± 0.473a 1.739 ± 0.156 0.097 ± 0.014 nd

1.289 ± 0.042 0.768 ± 0.032 0.066 ± 0.007 0.397 ± 0.020

0.408 ± 0.026 0.229 ± 0.017 0.072 ± 0.005 0.116 ± 0.009

0.167 ± 0.014 ndb nd 0.079 ± 0.002

0.068 ± 0.007 nd nd nd

Standard error of regression. Not determined.

Table 2 Estimated constants for predicting colour change (Eq. (5)), peptide formation (Eqs. (7) and (8)), and free amino group reduction (Eq. (10)) in reconstituted skim milk (10% w/ w) at different high pressure (p in MPa) and heat (T in °C) conditions using a reference pressure and temperature of 400 MPa and 110 °C, respectively. Constants

Estimated values

G0 G1 G2 G3 G4 G5

2.079 ± 0.181 (6.166 ± 1.086)  103 (5.305 ± 1.533)  102 (4.356 ± 0.573)  105 (1.036 ± 0.170)  107 (3.094 ± 1.198)  107

R2 = 0.966; Af = 1.268 F0 F1 F2

3.560 ± 0.099 (2.906 ± 0.339)  103 (8.756 ± 0.668)  102

R2 = 0.955; Af = 1.297 F3 F4 F5

(3.536 ± 0.068)  101 (2.279 ± 0.204)  104 (7.583 ± 0.450)  103

R2 = 0.981; Af = 1.146 H0 H1 H2 H3

6.522 ± 0.058 (1.438 ± 0.229)  103 (6.312 ± 0.511)  102 (6.720 ± 1.650)  105

R2 = 0.986; Af = 1.139

assumption is only hypothetical. Overall, high pressures accelerate the brown colour development of skim milk during heating. Maillard reaction in milk, from a quantitative point of view, is more dominant than lactose isomerisation/degradation when heating is carried out at a lower temperature (i.e., <100 °C); an opposite trend is found upon heating above 100 °C (van Boekel, 1998).

In our study, it is possible that the resultant coloured products from HPTP of skim milk come primarily from lactose degradation similar to the browning reaction in milk heat treated at atmospheric pressure. To best of our knowledge, there is no published data on the formation of brown pigment from lactose degradation in milk serum as affected by high pressure. Previously, Moreno et al. (2003) reported a decrease in browning of lactose 10% w/w in carbonatebicarbonate buffer (pH 10.0) during HPTP (400 MPa, 60 °C, 3 h). The decreased formation of brown colour of a lactose solution during HPTP was confirmed by the decrease of lactose conversion to lactulose from 18.8% (0.1 MPa at 60 °C for 3 h) to 7.8% (400 MPa at 60 °C for 3 h) of lactose and reduced galactose degradation from 12.6% to 1.8% of galactose under the same heat and pressure treatments. In the same study, HPTP was also applied to lactose 10% w/w in sodium hydroxide buffer (pH 10.2 and 10.6) which resulted in nearly no effects of pressure on browning. Due to the differences between prepared buffer systems and milk serum, it is difficult to relate the nature of lactose degradation during HPTP in this present study and how it affects the Maillard reaction. 3.3. Proteolysis In the present study, the proteolysis progress in milk during HPTP was followed by using a chromatographic method. Instead of selecting one or several peaks from the chromatographs, the resultant peptide peaks were grouped according to their polarities. In a reverse phase chromatographic separation, peptides that elute earlier are more polar than those eluting later. Therefore, the peptides were classified into hydrophilic (retention time (RT) of 11.60–27.15 min) and hydrophobic peptides (RT of 27.15–43.50 min). For modelling purposes, the total area (A) of hydrophilic and hydrophobic peptides was compared to the area at zero time

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

107

Fig. 3. Correlation between the experimental Z values and the predicted Z values for each modelling: (A) colour change, where Z is –Ln (G); (B) proteolysis (j for hydrophilic and h for hydrophobic peptide release), where Z is Ln (F); and (C) apparent reduction of free amino groups, where Z is Ln (H).

of treatment (A0), respectively, and normalised by deducting each index by factor one (A/A0 – 1). The area of each peptide group increased with increasing treatment time, which appears to follow particular patterns depending on the applied pressure and temperature. Previous studies by Hindle and Wheelock (1970) showed that proteolysis of milk, as confirmed by non-protein nitrogen content in acid-soluble fractions, starts at temperatures as low as 50 °C and accelerates with increasing temperature. The proteolysis of milk progresses linearly during the first 30 min of heating (Hindle and Wheelock, 1970; Gaucheron et al., 1999). Most studies on heat induced proteolysis in milk or caseinate systems indicate that heat induced proteolysis followed a zero order kinetic (Hindle and Wheelock, 1970; Hustinx et al., 1997; Morales and van Boekel, 1998). According to Metwalli and van Boekel (1998), a reaction appears to follow a zero order reaction when less than 30% of reactants have been converted into products. In order to have a better understanding on the order of reaction with respect to time, a conversion of 30–50% of reactants is required. Using a b-caseinate model system (1–5% w/w in simulated milk serum, pH 6.5), van Boekel (1999) followed hydrolysis during heating at 110–145 °C up to 125 min and observed a deviation from zero order reaction at higher degree protein hydrolysis. The order of reaction shifted from zero at 110 °C to 2 at 140 °C. The kinetic of proteolysis in skim milk during isobaric/isothermal treatments showed increased protein breakdown when the temperature (Fig. 4A and B) and/or pressure (Fig. 4C and D) was increased. Each kinetic curve was fitted to a power law model (Eq. (6)).



A A0



¼ 1 þ Ft x

ð6Þ

where F and x are the coefficient and the power value in the primary model of proteolysis, respectively, and t (min) is the treatment time. Power value x (Eq. (6)) was determined by testing values in the range of 0.44–1.84 in 0.01 increments for hydrophilic and hydrophobic peptide groups, respectively. The best (i.e., lowest) values P of the SE were found for power values of 0.96 and 0.78 for hydrophilic and hydrophobic peptide groups, respectively. The estimated parameters obtained from regression fitting of Eq. (6) are presented in Table 3. The data set was then used to describe the pressure and temperature dependence of F. A linear model (Eq. (7)) and a non-linear model (Eq. (8)) were selected to deliver a good description of F as a function of pressure and temperature for the formation of hydrophilic and hydrophobic peptides, respectively. The estimated parameters for Eqs. (7) and (8) using a reference pressure p0 = 400 MPa and temperature T0 = 110 °C are listed in Table 2.

Ln ðFÞ ¼ F 0 þ F 1 ðp  p0 Þ þ F 2 ðT  T 0 Þ

ð7Þ

1 F 3 þ F 4 ðp  p0 Þ þ F 5 ðT  T 0 Þ

ð8Þ

Ln ðFÞ ¼

with F0, F1, F2 and F3, F4, F5 are the coefficients in the secondary model for the hydrophilic and hydrophobic peptide group formation, respectively. The parity plot (Fig. 3B) illustrating the natural logarithm of the experimental versus the predicted F values indicates no significant heteroskedasticity for both peptide group formations as the deviations from the bisector are small. In addition, both data sets are correlated with R2 values of 0.955 and 0.981 for hydrophilic and hydrophobic peptide groups, respectively. Their Af values are 1.297 (hydrophilic) and 1.146 (hydrophobic). Substituting F in Eq. (6) with the pressure–temperature relations of Eqs. (7) and (8) for hydrophilic or hydrophobic peptide groups, respectively, allows the determination of pressure–temperature combinations that result in a targeted degree of peptide formation in skim milk, which is described as the increase of area under the peptide chromatograms. Using the temperature as an independent variable and a targeted degree of proteolysis (A/A0  1) at a given treatment time, the equation can be solved to obtain the pressure. Fig. 5 shows the calculated pressure– temperature combinations which give a specific degree of peptide formation after 30 min isothermal/isobaric treatment for either peptide group. For example, at atmospheric pressure, heating at 134.5 °C for 30 min is required to triple hydrophilic peptide formation ((A/A0 – 1) = 2), whereas HPTP at 300 MPa and 125 °C is predicted to give the same degree of peptide formation. In contrast, atmospheric heating at 125 °C results in a degree of hydrophilic peptide formation of approximately 0.9, implying that pressure increases hydrolysis of proteins and thus peptide formation in skim milk. 3.4. Reduction of free amino groups Reconstituted skim milk (10% w/w) showed faster reduction of free amino groups with increasing temperature (Fig. 6A). However, the decrease of free amino groups slowed with increasing pressures when the temperature was kept constant (Fig. 6B). The proteolysis evaluation showed that the formation of peptides increased with increasing temperatures. However, since free amino groups were reduced, it is likely that the exposed N-terminal a-amino groups of peptides or proteins quickly conjugated with lactose and/or its isomerisation/degradation products. Thus, it can be hypothesised that the rate of apparent lactose conjugation was faster than the rate of proteolysis, resulting in an overall reduction of

108

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

Fig. 4. Peptide release in reconstituted skim milk (10% w/w), as expressed by (A/A0)  1, during heating at 100–140 °C and 0.1 MPa for (A) hydrophilic and (B) hydrophobic peptide groups and during HPTP at 120 °C and 0.1–600 MPa for (C) hydrophilic and (D) hydrophobic peptide groups. Solid lines are obtained by fitting the experimental data to Eq. (6) with power values of 0.96 and 0.78 for hydrophilic and hydrophobic peptide groups, respectively.

Table 3 Estimated values of coefficient F in Eq. (6) for combined high pressure (p in MPa) and thermal (T in °C) induced peptide formation in reconstituted skim milk (10% w/w) with x equals to 0.96 and 0.78 for hydrophilic and hydrophobic peptide groups, respectively. p/T

a b

100

110

120

130

140

Hydrophilic peptide group 0.1 0.245 ± 0.034a 200 0.791 ± 0.022 400 1.635 ± 0.053 600 nd

0.534 ± 0.017 1.971 ± 0.056 2.774 ± 0.027 5.827 ± 0.085

2.390 ± 0.092 4.692 ± 0.077 7.326 ± 0.191 10.617 ± 0.152

5.899 ± 0.214 ndb nd 18.814 ± 0.502

14.251 ± 0.287 nd nd nd

Hydrophobic peptide group 0.1 0.502 ± 0.074 200 1.669 ± 0.053 400 2.590 ± 0.062 600 nd

1.635 ± 0.116 4.013 ± 0.120 4.728 ± 0.117 8.077 ± 0.216

6.050 ± 0.172 8.045 ± 0.151 10.041 ± 0.218 12.656 ± 2.693

9.624 ± 0.254 nd nd 18.678 ± 0.788

14.236 ± 0.614 nd nd nd

Standard error of regression. Not determined.

free amino acids during treatment at high temperatures. In contrast, the rate of C/C0 reduction was slower at higher pressures; this could refer to a slower conjugation rate between amino acids and sugars compared to the rate of proteolysis, leaving more free N-terminal a-amino groups of peptides to react with OPA. The highest browning, observed at high pressures, was a combined output of complex chemical and physical reaction, i.e., amino acids and sugars conjugation, sugar degradation, and casein micelle dissociation. The measurement of milk samples showed a large difference of lightness (DL) due to casein micelle dissociation and, thus, DL contributed the most to DEab according to Eq. (1). Therefore, a great colour change could be obtained if the lightness decreases remarkably although the conjugation rate was reduced.

In order to model the kinetics of the reduction of free amino acids, each kinetic curve was fitted to Eq. (9). The obtained values for H at the different pressure–temperature conditions tested are compiled in Table 4.

C ¼ expðHtÞ C0

ð9Þ

where H is the coefficient in the primary model for free amino acids reduction and t (min) is the treatment time. A secondary model (Eq. (10)), which describes H as a function of pressure and temperature was developed with reference condition of p0 = 400 MPa and T0 = 110 °C. The estimated parameters obtained from regression fitting of Eq. (10) are presented in Table 2.

109

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

Fig. 5. Predictive pressure–temperature diagrams of peptide formation in reconstituted skim milk (10% w/w) during HPTP for 30 min, with lines showing increase of (A/ A0)  1 by (A) 0.1–20 for hydrophilic and (B) 0.1–3 for hydrophobic peptide groups.

Fig. 6. Kinetics of free amino group reduction in reconstituted skim milk (10% w/w) during (A) thermal treatment at 100–140 °C and 0.1 MPa and (B) HPTP at 120 °C and 0.1–600 MPa. Solid lines are obtained by fitting the experimental data to Eq. (9).

Table 4 Estimated rate constants H (103) in Eq. (9) for combined high pressure (p in MPa) and heat (T in °C) induced free amino acids reduction in reconstituted skim milk (10% w/w). p/T 0.1 200 400 600 a b

100 a

1.013 ± 0.449 0.829 ± 2.831 0.892 ± 0.372 nd

110

120

130

140

2.292 ± 0.531 1.945 ± 0.357 1.554 ± 0.430 0.918 ± 0.342

9.068 ± 1.074 4.411 ± 0.184 2.812 ± 0.529 2.008 ± 0.493

13.798 ± 0.962 ndb nd 2.872 ± 0.625

36.671 ± 5.462 nd nd nd

Standard error of regression. Not determined.

The parity plot (Fig. 3C) confirms no significant heteroskedasticity during the mathematical modelling as the deviations from the bisector are small. The data and the model have R2 of 0.986 and Af of 1.139.

Ln H ¼ H0 þ H1 ðp  p0 Þ þ H2 ðT  T 0 Þ þ H3 ðp  p0 ÞðT  T 0 Þ

ð10Þ

where H0, H1, H2, and H3 are the coefficients in the secondary model of free amino acids reduction. Substituting Eq. (10) into Eq. (9) allows the construction of a pressure–temperature diagram showing combinations which results in a targeted C/C0 during HPTP for a fixed period of time. Increasing the pressure at constant temperature slowed the

reduction of C/C0; for example, heating at 113 °C at ambient pressure for 30 min results in a C/C0 of 0.9, whereas the same temperature at 430 MPa results in a C/C0 of 0.95 (data not shown). A study by Buckow et al. (2011) on glycosylation of BSA (2.5 mg/mL BSA and 25 mg/mL glucose in 0.05 M bicine buffer, pH 9) reported a hampered Maillard reaction during HPTP (70–132 °C and 0.1–600 MPa up to 2 h) as displayed by slower reduction of free amino acids. The reason beyond the decelerated glycosylation was not explicitly addressed due to the experimental design which focused on the observation of the Maillard reaction as a whole and disregarded the reaction stages. It suggested that the overall activation volume (DV*) for Maillard reaction between

110

A.F. Devi et al. / Journal of Food Engineering 147 (2015) 102–110

BSA and glucose was possibly positive, and thus, hampered under high pressure. 4. Conclusions HPTP increased colour change and proteolysis in reconstituted skim milk (10% w/w) compared to thermal treatments at atmospheric pressure. The rates of both reactions increased with increasing temperature and pressure. The apparent reduction of free amino acids was faster at higher temperatures; however, the reduction was slowed under high pressure. No reaction mechanism can be elucidated from this study because the observation was for the Maillard reaction as a whole, ignoring the complexity of intermediate stages. Further studies, involving mass spectroscopy, on each stage of the Maillard reaction using both skim milk and milk model systems will help understanding the mechanisms of Maillard reaction and to characterise the conjugates in skim milk during HPTP. Nevertheless, the knowledge and predictive models generated from this study might be beneficial for the development of high pressure assisted thermal sterilisation procedures for heat-sensitive, high-value dairy products including infant formula and concentrates. It is possibly also of interest to investigate the progress of the Maillard reaction during and after HPTP to valuable dairy products, bioactive peptides, or pharmaceuticals where unwanted reactions resulting in loss of nutrients, toxic compounds, off-flavours, and discolouration have to be avoided through strict process control. Acknowledgements Author A.F. Devi would like to express her gratitude to Endeavour International Postgraduate Research Scholarship for providing financial support during her study. Technical assistance of Dr. Roderick Williams, Mr. Phil Muller and Mr. Piotr Swiergon is very much appreciated. References Baranyi, J., Pin, C., Ross, T., 1999. Validating and comparing predictive models. Int. J. Food Microbiol. 48, 159–166. Buckow, R., Wendorff, J., Hemar, Y., 2011. Conjugation of bovine serum albumin and glucose under combined high pressure and heat. J. Agric. Food Chem. 59, 3915– 3923. Burton, H., 1994. Chemical and physical changes in milk at high temperatures. In: Ultra-High-Temperature Processing of Milk and Milk Products, Springer Science+Business Media, New York, pp. 44–76. Chevalier, F., Kelly, A.L., 2010. Proteomic quantification of disulfide-linked polymers in raw and heated bovine milk. J. Agric. Food Chem. 58, 7437–7444. Chevalier, F., Hirtz, C., Sommerer, N., Kelly, A.L., 2009. Use of reducing/nonreducing two-dimensional electrophoresis for the study of disulfide-mediated interactions between proteins in raw and heated bovine milk. J. Agric. Food Chem. 57, 5948–5955. de Wit, J.N., Klarenbeek, G., 1984. Effects of various heat treatments on structure and solubility of whey proteins. J. Dairy Sci. 67, 2701–2710. Devi, A.F., Buckow, R., Hemar, Y., Kasapis, S., 2013. Structuring dairy systems through high pressure processing. J. Food Eng. 114, 106–122. Fox, P.F., 1981. Heat stability of milk: significance of heat-induced acid formation in coagulation. Irish J. Food Sci. Technol. 5, 1–11. Gaucheron, F., Famelart, M., Mariette, F., Raulot, K., Michela, F., Le Graeta, Y., 1997. Combined effects of temperature and high-pressure treatments on physicochemical characteristics of skim milk. Food Chem. 59, 439–447.

Gaucheron, F., Mollé, D., Briard, V., Léonil, J., 1999. Identification of low molar mass peptides released during sterilization of milk. Int. Dairy J. 9, 515–521. Grauwet, T., Rauh, C., Van der Plancken, I., Vervoort, L., Hendrickx, M., Delgado, A., Van Loey, A., 2012. Potential and limitations of methods for temperature uniformity mapping in high pressure thermal processing. Trends Food Sci. Technol. 23, 97–110. Guo, M., Fox, P.F., Flynn, A., Mahammad, K.S., 1989. Heat-induced changes in sodium caseinate. J. Dairy Res. 56, 503–512. Heinz, V., Buckow, R., 2010. Food preservation by high pressure. Journal für Verbraucherschutz und Lebensmittelsicherheit – J. Consum. Prot. Food Saf. 5, 73–81. Hernández, M., Camañas, R., Alvarez-Coque, M., 1990. Determination of total free amino acids with o-phthalaldehyde and N-acetyl-L-cysteine. Microchem. J. 42, 288–293. Hindle, E., Wheelock, J., 1970. The release of peptides and glycopeptides by the action of heat on cow’s milk. J. Dairy Res. 37, 397–405. Huppertz, T., Kelly, A.L., Fox, P.F., 2002. Effects of high pressure on constituents and properties of milk. Int. Dairy J. 12, 561–572. Hustinx, J.C.A., Singh, T.K., Fox, P.F., 1997. Heat-induced hydrolysis of sodium caseinate. Int. Dairy J. 7, 207–212. Jaeger, H., Janositz, A., Knorr, D., 2010. The Maillard reaction and its control during food processing. The potential of emerging technologies. Pathol. Biol. 58, 207–213. Kessler, H.G., 1996. Food engineering and dairy technology, fourth ed. Verlag A. Kessler, Freising, Germany. Le, T., Deeth, H.C., Bhandari, B., Alewood, P.F., Holland, J.W., 2012. A proteomic approach to detect lactosylation and other chemical changes in stored milk protein concentrate. Food Chem. 132, 652–662. Lochmann, D., Stadlhofer, S., Weyermann, J., Zimmer, A., 2004. New protamine quantification method in microtiter plates using o-phthaldialdehyde/N-acetylL-cysteine reagent. Int. J. Pharm. 283, 11–17. Metwalli, A.A.M., van Boekel, M.A.J.S., 1998. On the kinetics of heat-induced deamidation and breakdown of caseinate. Food Chem. 61, 53–61. Morales, F.J., Jiménez-Pérez, S., 1998. Monitoring of heat-induced proteolysis in milk and milk-resembling systems. J. Agric. Food Chem. 46, 4391–4397. Morales, F.J., van Boekel, M.A.J.S., 1998. A study on advanced Maillard reaction in heated casein/sugar solutions: colour formation. Int. Dairy J. 8, 907–915. Morales, F.J., Romero, C., Jiménez-Pérez, S., 1995. New methodologies for kinetic study of 5-(hydroxymethyl)-furfural formation and reactive lysine blockage in heat-treated milk and model systems. J. Food Prot. 58, 310–315. Moreno, F.J., Villamiel, M., Olano, A., 2003. Effect of high pressure on isomerization and degradation of lactose in alkaline media. J. Agric. Food Chem. 51, 1894–1896. Nabhan, M.A., Girardet, J.M., Campagna, S., Gaillard, J.L., Le Roux, Y., 2004. Isolation and characterization of copolymers of beta-lactoglobulin, alpha-lactalbumin, kappa-casein, and alpha(s1)-casein generated by pressurization and thermal treatment of raw milk. J. Dairy Sci. 87, 3614–3622. O’Connell, J.E., Fox, P.F., 2003. Heat-induced coagulation of milk. In: Fox, P.F., McSweeney, P.L.H. (Eds.), Advanced Dairy Chemistry—1 Proteins. Springer, US, pp. 879–945. Orlien, V., Knudsen, J.C., Colon, M., Skibsted, L.H., 2006. Dynamics of casein micelles in skim milk during and after high pressure treatment. Food Chem. 98, 513– 521. Pagliarini, E., Vernile, M., Peri, C., 1990. Kinetic study on color changes in milk due to heat. J. Food Sci. 55, 1766–1767. Patel, H.A., Singh, H., Anema, S.G., Creamer, L.K., 2006. Effects of heat and high hydrostatic pressure treatments on disulfide bonding interchanges among the proteins in skim milk. J. Agric. Food Chem. 54, 3409–3420. Patel, H.A., Anema, S.G., Holroyd, S.E., Singh, H., Creamer, L.K., 2007. Methods to determine denaturation and aggregation of proteins in low-, medium- and high-heat skim milk powders. Lait 87, 251–268. Rizzi, G.P., 1997. Chemical structure of colored Maillard reaction products. Food Rev. Int. 13, 1–28. Schrader, K., Buchheim, W., 1998. High pressure effects on the colloidal calcium phosphate and the structural integrity of micellar casein in milk, II. Kinetics of the casein micelle disintegration and protein interactions in milk. Kieler Milchwirtschaftliche Forschungsberichte 50, 79–88. van Boekel, M.A.J.S., 1998. Effect of heating on Maillard reactions in milk. Food Chem. 62, 403–414. van Boekel, M.A.J.S., 1999. Heat-induced deamidation, dephosphorylation and breakdown of caseinate. Int. Dairy J. 9, 237–241.