Innovative Food Science and Emerging Technologies

Transcription

1 Innovative Food Science and Emerging Technologies 26 (2014) Contents lists available at ScienceDirect Innovative Food Science and Emerging Technologies journal homepage: Dynamic ultra-high pressure homogenisation of milk casein concentrates: Influence of casein content Hanne Sørensen a,b,, Kell Mortensen c, Geir Humborstad Sørland d, Flemming Hofmann Larsen a, Marie Paulsson e, Richard Ipsen a a Department of Food Science, University of Copenhagen, Rolighedsvej 30, 1958 Frederiksberg C, Denmark b Arla Foods amba, Denmark c Niels Bohr Institute, X-ray and Neutron Science, University of Copenhagen, Universitetsparken 5, D306, 2100 Copenhagen, Denmark d Anvendt Teknologi A/S, 7022 Trondheim, Norway e Department of Food Technology, Engineering and Nutrition, Lund University, P.O. BOX 124, SE Lund, Sweden article info abstract Article history: Received 17 June 2014 Accepted 18 September 2014 Available online 28 September 2014 Editor Proof Receive Date 9 October 2014 Keywords: Casein concentrate Process Ultra-high pressure homogenisation Microstructure Milk casein concentrates with different casein-to-protein ratios (0.82 and 0.93), ph (5.8, 6.2 and 6.7) and level of applied ultra-high pressure homogenisation (UHPH) (0, 150 and 300 MPa) were studied. Samples with a caseinto-protein ratio of 0.93 at ph 5.8 and subjected to UHPH treatment at 300 MPa had the highest apparent viscosity and particle size followed by similar samples with a casein-to-protein ratio of Water mobility was reduced at higher casein-to-protein ratio and with higher ph, while increased water mobility was observed in samples with a casein-to-protein ratio of 0.82 at ph 5.8 subjected to UHPH treatment at 300 MPa. The 1 Hand 31 PNMR spectra revealed minor structural effects of adding casino-phospho-peptides and changing ph, but no effect of UHPH treatment, which indicate that UHPH treatment did not induce measureable structural changes in the individual proteins but rather induced changes in the tertiary structure or degree of polymerisation. Industrial relevance: Milk depleted from whey protein is advantageous to use in cheese production since it can be further processed without detrimental denaturation of whey protein which will result in reduced maturation of cheese. In combination with ph control and ultra-high pressure homogenisation of the milk casein concentrate structural changes in the protein can be obtained and new interesting milk based products can be developed Elsevier Ltd. All rights reserved. 1. Introduction Processing of milk can induce structural changes of the milk proteins at both micro- and macro molecular levels (Kulozik, 2008). In this context ultra-high pressure homogenisation (UHPH) represents a purely physical and mechanical method for this purpose (Roach & Harte, 2008). Previously the effects of UHPH treatment on isolated systems of whey protein, caseins or on systems based on casein to whey protein ratio of 80:20 as found in bovine milk have been assessed (Fox & McSweeney, 1998). UHPH treatment induces denaturation and aggregation of whey proteins as well as changes in the casein micellar structure depending on e.g. chemical composition of the solution, velocity gradients, inlet and outlet temperature, holding time and especially the mechanical forces introduced by homogenisation (Dumay et al., 2013). Only few reports describe the effect of UHPH treatment applied to pure whey protein (Bouaouina et al., 2006; Grácia-Juliá et al., 2008) or Corresponding author at: Arla Foods amba, Arla Strategic Innovation Centre, Rørdrumvej 2, 8220 Brabrand, Denmark. Tel.: address: hsor@arlafoods.com (H. Sørensen). pure casein systems (Roach & Harte, 2008). Grácia-Juliá et al. (2008) studied the effect of UHPH treatment on whey protein isolate and observed protein aggregation at pressures exceeding 225 MPa (holding time b 1 s). They furthermore observed that the mechanical forces were more important than heating in relation to denaturation. Roach and Harte (2008) analysed a native casein suspension and reported reduced solubilisation of casein at 250 MPa and above. In addition constant concentrations of α-lactalbumin and β-lactoglobulin were observed. Other researchers studied the effects of UHPH treatment on protein denaturation in skim milk and observed less or no denaturation of whey proteins at pressures up to 200 MPa or at holding times less than 0.7 s (Hayes & Kelly, 2003; Pereda et al., 2009; Sandra & Dalgleish, 2005). However, denaturation of β-lg has been reported to occur at longer holding times (Datta et al., 2005; Hayes et al., 2005). Escobar et al. (2011) compared to cheeses prepared from raw milk and UHPH treated (300 MPa, 2 s) raw milk and found no evidence of denaturation of β-lg due to UHPH treatment. On the other hand, Zamora et al. (2007) reported reduced amounts of β-lg and α-la in cheese whey due to UHPH treatment, because of incorporation into the curd, and later Zamora et al. (2012) showed that UHPH treatment increased incorporation of both whey protein and caseins in the curd. The UHPH treatment procedures / 2014 Elsevier Ltd. All rights reserved.

2 144 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) in both of these studies were based on piston-gap homogenisers with an inlet temperature of 30 C, maximum homogenisation temperature of 90 C and a holding time 0.7 s. Since denaturation and solubilisation of proteins depend on storage conditions and temperature before and after UHPH treatment it is very difficult to make a direct comparison of the studies mentioned above as they adopt different experimental set-ups. Milk depleted from whey protein is advantageous to use in cheese production since it can be further processed without detrimental denaturation of whey protein which will result in reduced maturation of cheese due to interactions between β-lactoglobulin and κ-casein (Hyslop, 2003). For that reason milk casein concentrates with two different casein-to-protein ratios processed by dynamic UHPH treatment have been analysed with respect to apparent viscosity, particle size, soluble whey protein and using low as well as high-field NMR to assess the impact on micro- and macrostructure of the samples. 2. Materials and methods 2.1. Preparation of samples Raw milk (Arla Foods, Brabrand, Denmark) was skimmed (55 C) and skim milk (3000 L) was heated (63 C, 15 s) and cooled to 5 C (Pasilac Therm A/S, Silkeborg, Denmark). The following day skim milk was pre-heated to 55 C in 300 s (Pasilac Therm, Silkeborg, Denmark). The MF process was then carried out in batch mode at 50 C using a pilot unit (500 L h 1 ) constructed by Arla Foods (Arla Foods, Videbæk, Denmark), equipped with 24 pcs FR3B-6338 SW-membranes (Synder, Vacaville, CA, USA) with a nominal molecular mass cut-off of 800 kda and a UF unit (Arla Foods, Videbæk, Denmark) equipped with 16 pcs HFK SW-membranes (Koch, Wilmington, DE USA) with a nominal molecular mass cut-off of 5 kda. The skim milk was concentrated to 5% (w/w) casein and diafiltration was carried out using five volumes of UF permeate from the MF permeate (casein-to-protein 0.93, C:P93), while the other milk casein concentrate solution (casein-toprotein 0.82, C:P82) was standardised from C:P93 by adding UF concentrate and UF permeate. The two milk casein concentrates (MCC) were pasteurised (72 C, 15 s) and cooled to 8 C (Pasilac Therm A/S, Silkeborg, Denmark). From each fraction C:P93 and C:P82 respectively, six samples were prepared in an experimental design that was created using Modde 9.1 (Umetrics, Umeå, Sweden). Within each fraction two samples were acidified to ph 5.8, one sample to ph 6.2 and three samples were not acidified (ph 6.7). Acidification was performed with 20% (w/w) citric acid (Jungbunzlauer, Basel, Schweiz). The same amounts of liquid (water, citric acid or water and citric acid) were added to all samples to obtain the same dilution factor. One sample at ph = 5.8 and one at ph = 6.7 was UHPH treated at 300 MPa and one at ph = 6.2 was UHPH treated at 150 MPa (DEE International Inc., Debee 2000, Boston, MA, USA). The samples were cooled and stored at 5 C. All samples were analysed after 2 10 days. The experimental design is summarised in Table 1. Two groups of samples were prepared: one with a casein-toprotein ratio of 0.93 (C:P93) and one with 0.82 (C:P82) and for each MCC the ph was adjusted to 5.8, 6.2 or 6.7. Subsequently, samples at ph 5.8 and ph 6.7 were UHPH treated at 0 and 300 MPa, while samples at ph 6.2 were subjected to a pressure of 150 MPa. There is a difference in ph between the experimental design and the actual measured values of ph, because of equilibrium setting after acidification. All trials were performed in duplicate Chemical analyses The ph was measured using a PHM-240 ph-metre (Hach Lange, Brønshøj, Denmark). Total solid content, fat and nitrogen were determined according to IDF standard methods (IDF, 1987, 2004a, 2008). Protein was obtained by multiplying the nitrogen content by a Kjeldahl Table 1 Experimental design (1). The sample code, given by the abbreviations C:Px_y_z, where x is the casein-to-protein ratio, y is the ph and z is the pressure in MPa during ultra-high pressure homogenisation (UHPH). Sample code Casein-to-protein ratio C:P93_5.8_ C:P93_5.8_ C:P93_6.2_ C:P93_6.7_ C:P93_6.7_0_ C:P93_6.7_0_ C:P82_5.8_ C:P82_5.8_ C:P82_6.2_ C:P82_6.7_ C:P82_6.7_0_ C:P82_6.7_0_ factor of 6.38 for milk protein, while the casein content was determined by multiplying the nitrogen content by 6.36 (Van Boekel & Ribadeau-Dumas, 1987). Non-casein nitrogen was determined according to the IDF standard method (IDF, 2004b) and ash content was determined according to NMKL standard method (NMKL, 2005). Determination of lactose was done using a lactose/day-galactose enzymatic Boehringer Mannheim test-kit (Roche, Basel, Schweiz). Total content of magnesium, sodium, chloride, potassium and phosphate was determined using inductively coupled plasma (ICP) spectroscopy (Perkin- Elmer Optima 4300 DV, Boston, MA, USA) Total calcium and serum calcium The total calcium was determined using inductively coupled plasma (ICP) optical emission spectrometry (ICP-OES) (Perkin-Elmer Opima 4300 DV, Boston, MA, USA). Samples of 0.5 g were weighed to a quartz tube and 5 ml of concentrated 69% (w/w) HNO 3 (SCP Science, Quebec, Canada) was added, and the sample was digested in a microwave (Anton Paar Multiwave 3000, Graz, Austria) and further diluted with ultrapure water. The serum calcium content was released from the MCC with centrifugation and filtration and the calcium in the serum was determined by ICP-OES: 0 30 ml of sample was centrifuged at 17,090 g, at 4 C for 2 h (Beckman Coulter Inc., Brea, CA, USA). The sample was filtered through a pleated filer (Grade 580, retention: N20 μm, Frisenette ApS, Knebel, Denmark) and the filtrate, was centrifuged at 126,000 g, at 4 C for 1 h (Bechman coulter optima L-80 XP ultracentrifuge, Birkerød, Denmark). Subsequently samples were filtered through a 3 kda cutoff filter (Sartorius Stedim biotech, Göettingen, Germany) and centrifuged at 5170 g at 4 C for 16 h Total whey protein and soluble whey protein To determine the total whey protein content 200 μl of sample was transferred into an Eppendorf tube. One ml of reduction buffer (100 mm sodium citrate; 6 M Urea) (Merck KGaA, Darmstadt, Germany) and 20 μl of 1 M 1.4-dithioerythritol (DTE) (Sigma-Aldrich Chemie GmbH, Steinheim, Germany) were added to the sample and mixed. The sample was incubated for 60 min at 37 C, and centrifuged at 9300 g, at 5 C in 10 min, 200 μl of the supernatant was transferred to a vial and analysed with a high-performance liquid chromatography (HPLC) system (Agilent Technologies 1290, Santa Clara, CA, USA) equipped with a BioSuite C18 PA-B column (C 18,3.5μm; mm, Waters, Milford, MA, USA) and a UV-detector to detect whey protein at a wavelength of 214 nm (Agilent Technology, Santa Clara, CA, USA). Mobile phase A was 0.1% trifluoroacetic acid (TFA) (Fluka, St. Louis, MO, USA) in water and mobile phase B was 0.1% ph UHPH treatment (MPa) (1) All treatments were performed on milk casein concentrate based on skim milk. All experiments were performed in duplicate.

3 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) TFA in acetonitrile (MeCN) (Rathburn Chemicals Ltd., Walkerburn, Scotland) and a gradient from 33.2% 44.8% mobile phase B was used. The column temperature was set to 45 C, injection volume of sample was 5 μl, flow rate was 0.35 ml min 1 and run time was 16 min. The amount of soluble whey protein and casein content in the sample was also determined by HPLC. Twenty ml of sample was adjusted to ph using 1 M HCl (VWR International S.A.S., Fontenaysous-Bois, France). The sample was centrifuged at 17,090 g at 4 C in 10 min to separate the supernatant from the pellet. Subsequently the supernatant was centrifuged at 9300 g at 5 C in 10 min and transferred to a vial and analysed by HPLC (Agilent Technologies 1290, Santa Clara, USA) equipped with a Xbridge BEH 300 C18 column (C μm; 2.1 mm 250 mm, Waters, Milford, USA) and a UV-detector to detect soluble whey protein at a wavelength of 214 nm. Mobile phase A was 0.1% trifluoroacetic acid (TFA) in water and mobile phase B was 0.1% TFA in acetonitrile (MeCN) and a gradient from 0% 51% mobile phase B was used. The column temperature was set to 40 C, injection volume of sample was 20 μl, flow rate was 0.3 ml min 1 and run time was 40 min. The proteins were separated based on hydrophobicity and size, with the hydrophilic small peptides coming out first and the hydrophobic coming out as the last. A protein or a protein fragment was tentatively assigned to each retention interval in the chromatogram, even though different proteins or protein fragments may be present in the same retention interval. The amount of soluble whey protein was detected by another HPLC ethod. The supernatant (described in the section above, soluble whey protein) was analysed by HPLC (Agilent Technologies 1290, Santa Clara, USA) equipped with an Xbridge BEH 300 C18 column (C μm; 2.1 mm 250 mm, Waters, Milford, USA) and a UV-detector to detect soluble whey protein at a wavelength of 214 nm. Mobile phases A and B were used and a gradient from 17 to 44% mobile phase B (from 2 14 min) was used. The column temperature was set to 55 C, injection volume of sample was 20 μl for C:P93 samples and 5 μl for C:P82 samples, flow rate was 0.35 ml min 1 and total run time was 17 min Rheological measurements Flow curves of MCC were obtained using a controlled stress rheometer (AR 2000, TA Instruments Ltd., New Castle, USA). The measuring system was double concentric cylinders (rotor outer radius mm, rotor inner radius mm) and a double gap cup. Measurements were performed in the shear rate interval 1 to 800 s 1. Shear rates were logarithmically spaced with five data point per decade and a steady state of 1 min was applied for every data point. The MCC was temperature equilibrated for 1 h at 25 C prior to measurement. All measurements were performed at 25 C in duplicate Particle size distribution Volume- and number-based particle size distributions of milk concentrate isolate were obtained with a Master Sizer 3000 (Malvern Instruments Ltd., Worcestershire, UK). De-stilled, degassed water was used as dispersant. The refractive index was set to for particles and to 1.33 for the dispersant. Milk concentrate isolate was added into the dispersant in the dispersion unit until a laser obscuration between 5 10% was reached and the particle size distributions were measured. All measurements were performed in duplicate Quantitative T 1 and T 2 -measurements The two-dimensional low-field nuclear magnetic resonance (2D LF- NMR) measurements were performed using a pulsed 1 H-NMR Analyser (Maran Bench Top, Oxford Instruments Ltd., Witney, UK) with a magnetic field strength of 0.47 T, corresponding to a resonance frequency of 23.2 MHz for 1 H. The instrument was equipped with an 18-mm (o.d.) variable temperature probe. Samples of 3.0 g concentrate were used, and were temperature equilibrated for 0.5 h at 15 C before measurement. All measurements were performed in duplicate at 15 C. The 1 H data was recorded using the spoiler recovery sequence (Sørland et al., 2011) for fast simultaneous acquisition of longitudinal (T 1 ) and transverse (T 2 ) relaxation data. The relaxation data were extracted assuming a multi-exponential decay (Cohen & Mendelson, 1982) using Eq. (1) In ð Þ ¼ X i ρ i 1 exp SRD T i 1 exp 2nτ T i 2 n ½1; 2; 3; number of echoes acquiredš where I(n) is the intensity, ρ i is the fraction of component i, i being the index of component describing the system and T 1 i and T 2 i are the corresponding relaxation times. SRD is the variable Spoiler Recovery Delay, 2τ is the inter echo spacing, and n denotes the echo number. The experiment was performed with a τ value (time between 90 and 180 pulses) of 150 μs, 4 scans, 2048 echo numbers, GS1 = 4 G cm 1 duration of 5 ms, GS2 = 13 G cm 1 duration of 1 ms and SRD from s up to 5.0 s. The fitting was performed using the Anahess algorithm (Ukkelberg et al., 2010) High-field nuclear magnetic resonance Samples for high-field liquid state NMR were prepared by mixing of 495 μl of sample with 55 μl ofd 2 O (99% isotopic enrichment, Sigma- Aldrich, St. Louis, MO, USA) containing 5.8 mm TSP-d 4 (sodium salt of 3-(trimethylsilyl)propionic-2,2,3,4-d 4 acid, Sigma-Aldrich, St. Louis, MO, USA). Liquid-state NMR experiments were carried out using a Bruker Avance DRX-500 (11.7T) spectrometer (Bruker Biospin, Rheinstetten, Germany) operating at Larmor frequencies of and MHz for 1 Hand 31 P, respectively. All experiments were conducted at a sample temperature of 6 C using a double-tuned BBI (Broad Band Inverse detection) probe equipped for 5 mm (o.d.) NMR tubes. The 1 H experiments were performed using the zgcppr pulse sequence (pre-saturation followed by a composite 90 pulse) employing a recycle delay of 5 s, 64 scans, a spectral width of 10 khz and an acquisition time of s. The 31 P NMR experiments were carried out using the zgpg30 pulsesequence(a30 31 P pulse using 1 H decoupling during acquisition) utilizing a recycle delay of 2 s, 1024 scans, a spectra width of 10,162 Hz and an acquisition time of s during which 1 H decoupling (waltz16) was applied. Chemical shifts were referenced to TSP-d 4 at 0.0 ppm (internal reference) for the 1 H NMR spectra and to 85% H 3 PO 4 (external reference, Sigma-Aldrich, St. Louis, MO, USA) for the 31 P NMR spectra Statistical analysis Statistical design of the experiments was performed with Modde 9.1 (Umetrics, Malmö, Sweden), Statistical analyses of experimental data were carried out using one way ANOVA in Minitab 16 (Minitab Ltd., Coventry, United Kingdom). 3. Results and discussion 3.1. Characterisation of casein concentrate The chemical composition of the milk casein concentrates with two levels of whey protein is presented in Table 2.Significantly higher levels of Na, Mg, K, Ca and P and lower levels of non-casein nitrogen were observed in the sample with the higher casein protein ratio (0.93) due to a higher content of SerP and metal binding sites in the caseins compared to the whey proteins. In Tables 3 and 4 the chemical composition of the different samples are presented and no significant differences were ð1þ

4 146 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) Table 2 Chemical composition of fractionated microfiltrated milk concentrate with a casein-toprotein ratio of 0.93 (C:P93) and 0.82 (C:P82) (1). Component Casein-to-protein ratio 0.93 found for dry matter or total protein content (Table 3), while the measured mineral content depended on starting material and acidification (Table 4). Analysis of the data presented in Tables 3 and 4 showed a division of the samples into six groups, based on casein protein level, ph and minerals, while no effect was observed for ultra-high pressure homogenisation. Within each group with a casein-to-protein ratio of, respectively, 0.93 or 0.82, the sample with ph 5.8, 6.2 and 6.7 was grouped together Total whey protein, soluble and non-soluble proteins Casein-to-protein ratio 0.82 Dry matter (g 100 g 1 ) ± ± 0.02 Protein (g 100 g 1 ) 4.89 ± ± 0.02 Casein (g 100 g 1 ) 4.53 ± 0.02 a 3.98 ± 0.02 b Non-casein nitrogen (g 100 g 1 ) 0.06 ± 0.01 a 0.14 ± 0.02 b Non-protein nitrogen (g 100 g 1 ) ± ± Fat (g 100 g 1 ) 0.09 ± ± 0.01 Lactose (g 100 g 1 ) 4.02 ± ± 0.02 Ash (g 100 g 1 ) 0.89 ± 0.01 a 0.82 ± 0.01 b ph 6.77 ± 0.01 a 6.71 ± 0.01 b Calcium (mm) ± 0.00 a ± 0.00 b Phosphorous (mm) ± 0.00 a ± 0.00 b Magnesium (mm) 5.17 ± 0.00 a 4.91 ± 0.00 b Potassium (mm) ± 0.00 a ± 0.00 b Sodium (mm) ± 0.00 a ± 0.00 b Chloride (mm) ± 0.00 a ± 0.00 b (1) Parameters marked with a superscript letter are significantly different (P b 0.05) from each other. Dry matter, protein, casein, non-casein nitrogen, non-protein nitrogen, fat, lactose, ash are presented in percent (g 100 g 1 ); calcium, phosphorous, magnesium, potassium, sodium and chloride are presented in mm. The soluble proteins from C:P93 and C:P82 were suggested (Bonfatti et al., 2008) to be small peptides, medium peptides, β-casein fragment (hydrolysed from β-casein with the natural milk enzyme plasmin), α- lactalbumin, casein, unassigned fraction, β-lactoglobulin A and β- lactoglobulin B, and at ph 4.6 they were characterised in relation to skim milk and presented (Table 5). Based on mass spectrometry the unassigned fraction most likely consists of casein fragments (data not shown). Samples, with values lower than 100% indicate, indicate that the sample contain less than in skim milk. Table 3 Chemical composition of milk casein concentrate (1). The sample code, given by the abbreviations C:Px_y_z, where x is the casein-to-protein ratio, y is the ph and z is the pressure in MPa during ultra-high pressure homogenisation (UHPH). Sample code Dry matter (g 100 g 1 ) Protein (g 100 g 1 ) C:P93_5.8_ ± ± a C:P93_5.8_ ± ± b C:P93_6.2_ ± ± c C:P93_6.7_ ± ± d C:P93_6.7_0_ ± ± d C:P93_6.7_0_ ± ± d C:P82_5.8_ ± ± a C:P82_5.8_300 n.a. (2) n.a. (2) 6.06 b C:P82_6.2_ ± ± c C:P82_6.7_ ± ± d C:P82_6.7_0_ ± ± d C:P82_6.7_0_ ± ± d (1) Parameters marked with a superscript letter are significantly different (P b 0.05) from each other. Dry matter and protein are presented in percent (g 100 g 1 ). (2) n.a.; data not available. ph Table 4 Mineral composition of milk casein concentrate (1)(2). The sample code, given by the abbreviations C:Px_y_z, where x is the casein-to-protein ratio, y is the ph and z is the pressure in MPa during the ultra-high pressure homogenisation (UHPH). Sample code Ca-T (3) (mm) Ca-S (3) (mm) P-T (3) (mm) P-S (3) (mm) Mg-T (3) (mm) Mg-S (3) (mm) K-T (3) (mm) K-S (3) (mm) Na-T (3) (mm) Na-S (3) (mm) C:P93_5.8_ ± 0.27 a ± 0.14 a ± 0.43 a ± 0.25 a 5.89 ± 0.05 a 3.77 ± 0.01 a ± ± 0.23 a ± ± 0.12 a C:P93_5.8_ ± 0.27 a ± 0.48 a ± 0.43 a ± 0.13 a 5.82 ± 0.05 a 3.66 ± 0.06 a ± ± 0.47 a ± ± 0.08 ab C:P93_6.2_ ± 0.27 a ± 0.18 b ± 0.43 a ± 0.15 b 5.92 ± 0.05 a 3.17 ± 0.05 b ± ± 0.34 ab ± ± 0.29 ab C:P93_6.7_ ± 0.27 a 7.20 ± 0.03 c ± 0.43 a ± 0.21 c 5.92 ± 0.05 a 2.52 ± 0.02 c ± ± 0.36 bc ± ± 0.11 bc C:P93_6.7_0_ ± 0.27 a 6.82 ± 0.28 c ± 0.43 a ± 0.38 c 5.91 ± 0.05 a 2.49 ± 0.06 c ± ± 0.12 c ± ± 0.03 c C:P93_6.7_0_ ± 0.27 a 6.69 ± 0.33 c ± 0.43 a ± 0.03 c 5.80 ± 0.05 a 2.44 ± 0.10 c ± ± 0.51 c ± ± 0.21 c C:P82_5.8_ ± 0.45 b ± 0.05 a ± 0.38 b ± 0.05 a 5.68 ± 0.05 b 3.51 ± 0.01 a ± ± 0.01 a ± ± 0.01 a C:P82_5.8_ ± 0.45 b ± 0.04 a ± 0.38 b ± 0.34 a 5.58 ± 0.05 b 3.54 ± 0.01 a ± ± 0.02 a ± ± 0.02 a C:P82_6.2_ ± 0.45 b ± 0.06 b ± 0.38 b ± 0.09 b 5.66 ± 0.05 b 2.96 ± 0.00 b ± ± 0.01 b ± ± 0.08 b C:P82_6.7_ ± 0.45 b 6.73 ± 0.18 c ± 0.38 b ± 0.08 c 5.64 ± 0.05 b 2.34 ± 0.06 c ± ± 0.23 c ± ± 0.02 cd C:P82_6.7_0_ ± 0.45 b 6.77 ± 0.02 c ± 0.38 b ± 0.29 c 5.57 ± 0.05 b 2.38 ± 0.03 c ± ± 0.05 c ± ± 0.01 c C:P82_6.7_0_ ± 0.45 b 6.79 ± 0.43 c ± 0.38 b ± 0.39 c 5.65 ± 0.05 b 2.35 ± 0.09 c ± ± 0.08 c ± ± 0.02 d Calcium, phosphorous, magnesium, potassium and sodium are presented in mm. (1) Parameters marked with different superscript letters are significantly different (P b 0.05) from each other. Total mineral level was compared between C:P93 and C:P82, while serum mineral level was compared within the same group, respectively C:P93 and C:P82. (2) Ca-T; Calcium-Total, Ca-S; Calcium-Serum, P-T; Phosphorous-Total, P-S; Phosphorous-Serum, Mg-T; Magnesium-Total, Mg-S; Magnesium-Serum, K-T; Potassium-Total, K-S; Potassium-Serum, Na-T; Sodium-Total, Na-S; Sodium-Serum. (3)

5 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) Table 5 Content of soluble proteins at ph 4.6 (1)(2)(3)(4). The sample code, given by the abbreviations C:Px_y_z, where x is the casein-to-protein ratio, y is the ph and z is the pressure in MPa during ultra-high pressure homogenisation (UHPH). Retention interval Sample code Small peptides Medium peptides β-casein fragment α-la Casein Unassigned fraction β-lg B C:P93_5.8_ ± ± ± ± 0.0 b 21.5 ± 1.6 b ± ± 1.2 b 3.2 ± 2.4 ab C:P93_5.8_ ± ± ± ± 3.1 a 34.7 ± 3.4 a ± ± 0.9 a 3.6 ± 0.7 a C:P93_6.2_ ± ± ± ± 1.2 ab 27.3 ± 0.5 ab ± ± 0.0 b 1.7 ± 0.1 b C:P93_6.7_ ± ± ± ± 0.9 ab 29.6 ± 1.6 a ± ± 0.1 b 1.6 ± 0.1 b C:P93_6.7_0_ ± ± ± ± 0.6 ab 28.8 ± 1.5 ab ± ± 0.1 b 1.6 ± 0.1 b C:P93_6.7_0_ ± ± ± ± 2.1 ab 27.8 ± 3.4 ab ± ± 0.1 b 1.6 ± 0.2 b C:P82_5.8_ ± ± ± ± ± ± ± 6.4 C:P82_5.8_ ± ± ± ± ± ± ± 17.7 C:P82_6.2_ ± ± ± ± ± ± ± 7.8 C:P82_6.7_ ± ± ± ± ± ± ± 1.8 C:P82_6.7_0_ ± ± ± ± ± ± ± 8.6 C:P82_6.7_0_ ± ± ± ± ± ± ± 8.0 (1) Parameters marked with different superscript letters are significantly different (P b 0.05) from each other. (2) Soluble proteins at ph 4.6; calculated as total area for each component divided by protein content in the sample/total area for the same component in skim milk divided by the protein content in skim milk. (3) Small peptides, medium peptides, β-casein fragment, α-lactalbumin, casein, un-assigned fraction, β-lactoglobulin A, β-lactoglobulin B. (4) At each retention interval in the chromatogram a suggested protein or fragment were assigned, but in the same retention interval different proteins could be present. β-lg A The samples with low levels of whey protein (C:P93) could be divided into three groups; ph 5.8 with UHPH treatment at 300 MPa, ph 5.8 without UHPH treatment and all other samples (Table 5). In corroboration of this, a significantly higher level of soluble proteins was observed in samples with low ph (ph 5.8) and subjected to UHPH treatment (300 MPa) in each retention interval. No significant difference in soluble proteins was observed at ph 6.7 due to ultra-high pressure homogenisation, but in general, the soluble protein values (Table 5) were higher for the samples treated with UHPH. There was no significant difference in samples with the high whey protein content (C:P82), but at the same ph the soluble protein was generally observed to be increased in UHPH treated samples. At low ph 5.8 UHPH treatment strengthens the effect of casein solubilisation, increasing the amount of protein fragments in the serum phase and hence increasing the apparent viscosity of the milk casein concentrate. The total whey protein content was found to be 6.4% and 77.0% with respect to skim milk for C:P93 and C:P82, respectively (Table 6, second column). The distribution of whey proteins between serum and casein micelles will depend on ph, and more whey protein will be associated with the casein micelle at decreased ph (Law & Leaver, 2000). Data from Table 6 showed that the remaining part of the whey proteins was virtually non-soluble in the C:P93 samples, whereas a higher solubility for α-la than for β-lg was observed for the C:P82 samples. A significant lower soluble fraction of α-la, β-lg B and A was found in the high whey protein system (C:P82) samples at ph 5.8 UHPH treated at 300 MPa compared to samples not subjected to pressure homogenisation, while in the low whey protein system (C:P93) samples a general higher solubilisation of α-la, β-lg B and A was observed (Table 6). In the C:P82 solutions a significant denaturation (41.69% versus 26.07%, Table 6, fourth column) of the whey protein at low ph (ph 5.8) by pressure homogenisation was observed compared to samples at the same ph without pressure homogenisation. The solubilisation of proteins was higher for the samples with high whey protein content (C:P82) and a significant whey protein denaturation Table 6 Content of total whey protein, soluble whey protein and calculation of non-soluble whey protein (1). The sample code, given by the abbreviations C:Px_y_z, where x is the casein-to-protein ratio, y is the ph and z is pressure in MPa during the ultra-high pressure homogenisation (UHPH). Retention interval Total WP (2) α-la (3) Sample code Soluble (6) (a.u.) (8) β-lg B (4) Non-soluble (7) β-lg A (5) (a.u.) (8) (a.u.) (8) Soluble (6) Non-soluble (7) Soluble (6) C:P93_5.8_0 6.4 ± ± 0.02 c ± 0.00 b ± 0.00 b C:P93_5.8_ ± 0.04 a ± 0.00 a ± 0.00 a C:P93_6.2_ ± 0.02 bc ± 0.00 b ± 0.00 b C:P93_6.7_ ± 0.01 b ± 0.00 b ± 0.00 b C:P93_6.7_0_ ± 0.01 b ± 0.00 b ± 0.00 b C:P93_6.7_0_ ± 0.01 b ± 0.00 b ± 0.00 b C:P82_5.8_ ± ± 2.75 a ± 3.75 a ± 5.49 a C:P82_5.8_ ± 7.21 b ± 3.45 bc ± 2.67 b C:P82_6.2_ ± 3.70 ab ± 2.81 b ± 2.26 ab C:P82_6.7_ ± 0.84 ab ± 0.89 c ± 1.15 b C:P82_6.7_0_ ± 0.07 ab ± 0.10 bc ± 0.04 b C:P82_6.7_0_ ± 1.95 ab ± 0.51 bc ± 0.69 b Non-soluble (7) (1) Parameters marked with different superscript letters are significantly different (P b 0.05) from each other. (2) Total whey protein (WP); calculated as WP/Total area (TA) for sample divided by WP/TA for skim milk times 100%. (3) α-la; α-lactalbumin, soluble and calculated non-soluble α-la in % of total whey protein. (4) β-lg B; β-lactoglobulin B, soluble and calculated non-soluble β-lg B in % of total whey protein. (5) β-lg A; β-lactoglobulin A, soluble and calculated non-soluble β-lg A in % of total whey protein. (6) Soluble; calculated as WP (total whey protein in the sample) times (total area component divided by protein content and injection volume) divided by (total area for the same component in skim milk divided by protein content and injection volume). (7) Non-soluble; calculated as 100 minus soluble. No standard deviation because it is a calculated value. (8) a.u.; absorbance at 214 nm (arbitrary unit).

6 148 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) (less soluble whey protein) was measured at low ph (ph 5.8) UHPH treated at 300 MPa compared with no UHPH treatment. At higher whey protein concentrations whey proteins exist both as free proteins as well as in whey-casein complexes. Contrary to our results, where no effect of UHPH treatment was observed at ph 6.7, Roach and Harte (2008) found solubilisation of casein at 250 MPa and above and constant concentrations of α-la and β-lg in native casein suspensions based on nano-suspension (b30 kda) and micro-suspension (b0.2 μm) with ph re-adjusted to 6.7. When investigating UHPH treatment of skim milk at natural ph (ph ), little or no denaturation of whey protein has been observed using up to 200 MPa compared to high-pasteurised milk (90 C, 15 s) (Hayes & Kelly, 2003; Pereda et al., 2009). Escobar et al. (2011) demonstrated that raw milk and UHPH treated (300 MPa) raw milk exhibited similar β-lg bands in sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), while denaturation of β-lg was reported with holding times up to 20 s (Datta et al., 2005; Hayes et al., 2005). Our results show no significant denaturation of whey protein at ph 6.7 in UHPH treated (300 MPa) samples C:P82 and C:P93 with an estimated holding time of s at 300 MPa, but an increased solubilisation of proteins were measured (Table 6), which is in agreement with work done by Sandra and Dalgleish (2005). Zamora et al. (2007) reported reduced amount of β-lg and α-la in the cheese whey due to UHPH treatment, because of incorporation of proteins into the cheese curd. Zamora et al. (2012) showed increased incorporation in the curd of both whey protein and caseins due to UHPH treatment (300 MPa, 0.7 s (piston-gap)). The different reported holding times, shear times and ph in the UHPH unit can explain the different reported degrees of denaturation and solubilisation of proteins Rheological measurements Apparent viscosity against shear rate is shown in Fig. 1a and b. Depending on the treatment, milk casein concentrate samples behaved either as Newtonian fluids or exhibited shear thinning. The samples at ph 5.8 UHPH treated at 300 MPa showed shear thinning, while all the other samples showed Newtonian behaviour over the range of shear rates investigated. The apparent viscosities were significantly increased in samples at ph 5.8 subjected to UHPH treatment at 300 MPa, and no significant differences were observed between the other samples. The samples with less whey protein had the highest apparent viscosity, due to casein binding more water than native whey protein and this in accordance with previous results (Sørensen et al., 2013). Furthermore the more pronounced solubilisation of casein in samples with high casein level can likewise contribute to an increased apparent viscosity in the sample due to the increased apparent viscosity of the serum phase. The ratio between serum and total calcium and phosphorous was lower for the samples at ph 5.8 with high casein level compared to the same samples with low casein level, and this may have an, as yet, unexplainable effect on the apparent viscosity and thereby structure. At ph 6.7 no difference in apparent viscosity was seen when UHPH treatment was applied, in agreement with Sørensen et al. (2013). A decrease in apparent viscosity has, however, been observed in samples made from powder, when subjected to UHPH treatment at ph 6.7 (Chevalier-Lucia et al., 2011). It would thus appear that the effect of UHPH treatment on the apparent viscosity of milk casein concentrate depends on whether the starting material is based on milk or rehydrated powder. The structure of the proteins will be affected by the drying process due to the heating (Singh, 2007) Particle size distribution Data obtained from static light scattering showing volume based size distribution is presented in Fig. 2. for samples C:P93_5.8_300 and C:P82_5.8_300 (Fig. 2a) and for the remaining samples (Fig. 2b and c). The particle size for samples treated at ph 5.8 and UHPH treated Fig. 1. a. Apparent viscosity (Pa s) as function of shear rate (s 1 ) for sample, with low level of whey protein, at ph 5.8 without UHPH treatment ( ), with 300 MPa UHPH treatment ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). b. Apparent viscosity (Pa s) as function of shear rate (s 1 ) for sample, with high level of whey protein, at ph 5.8 without UHPH treatment ( ), with 300 MPa UHPH treatment ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). at 300 MPa was significantly larger than for the other samples. Sample C:P93_5.8_300 and C:P82_5.8_300 had particle sizes in the region of 4 to 100 μm and the other samples had particle sizes of 0.01 to 0.5 μm. Only at ph 5.8 and UHPH treatment at 300 MPa large aggregates were formed with a concomitant increase in apparent viscosity. This could be due to aggregation of disrupted casein micelles, casein fractions and whey protein, to decrease exposure of hydrophobic sites following the disruption caused by UHPH treatment. When the ph is decreased more calcium, phosphate and caseins are liberated into the soluble fraction (Gaucheron, 2005) and this makes the caseins more sensitive to changes, probably because of less interaction between the individual caseins. Casein aggregation (re-association of micellar fragments) at neutral ph, after isostatic pressure treatments at MPa has previously been reported (Huppertz & de Kruif, 2006; Huppertz et al., 2006b; 2006a) Low-field nuclear magnetic resonance The longitudinal (T 1 )andtransverse(t 2 ) relaxation times for a range of samples are plotted in Fig. 3. The relaxation times (T 1 ~ s, T 2 ~ s) were reduced significantly compared to the values of pure water (T 1 =T 2 ~ 2.7 s), indicating restricted mobility of water due the interaction with the protein network.

7 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) Fig. 3. Longitudinal (T 1 ) and transverse relaxation times (T 2 ) for samples with low whey protein content at ph 5.8 without UHPH treatment ( ), with 300 MPa UHPH treatment ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). Sample with high level of whey protein, at ph 5.8 without UHPH treatment ( ), with 300 MPa UHPH treatment ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). Fig. 2. a. Particle size distribution in volume for samples with low whey protein content at ph 5.8 with 300 MPa UHPH treatment ( ) and samples with high protein content at ph 5.8 with 300 MPa UHPH treatment ( ). b. Particle size distribution in volume for samples with low whey protein content at ph 5.8 without UHPH ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). c. Particle size distribution in volume for samples with high whey protein content at ph 5.8 without UHPH treatment ( ), at ph 6.2 with 150 MPa UHPH treatment ( ), at ph 6.7 with 300 MPa UHPH treatment ( ) and without UHPH treatment ( ). Asignificant lower mobility of water was observed in samples with low whey protein content (C:P93) compared to samples with high whey protein content (C:P82). This is primarily expressed in lower T 2 relaxation times. In addition higher mobility was observed for samples with lower ph and only for sample with high whey protein content (C:P82) at ph 5.8 a higher mobility was observed as resulting from UHPH treatment. The difference in water mobility between C:P93 and C:P82 is related to the difference in water binding capacity for the different proteins. Casein, whey protein and denatured whey protein bind, respectively, around 2 4, 0.5 and 10 g water per g protein (Robin et al., 1993). Hydration of proteins in general is related to the amino acid sequence (the charged sites are more hydrated than the nonionic polar sites), conformation, ph, ions, temperature and surface area (Kinsella & Fox, 1986). At the same ph, the C:P93 solution was expected to have more charged sites (Walstra & Jenness, 1983) than C:P82, due to the higher casein content and concentration of cations and therefore an expected better water binding. Phosphocaseinate studied with Carr Purcell Meiboom Gill (CPMG) NMR relaxometry, showed a decreased water relaxation rate to sample added lactose and a non-significant effect of added whey protein (Le Dean et al., 2001). However, our results showed an increased relaxation time in samples with whey protein due to UHPH treatment, which probably was due to denaturation of whey protein. The observed increased relaxation times at low ph have also been reported using 1-dimensional NMR (Mariette et al., 1993; Roefs et al., 1989). Mariette et al. (1993) explained the effect by release of calcium phosphate into the soluble phase. A more detailed explanation could be that less water will be in contact with protein at ph 5.8 compared to 6.7, due to reduced voluminosity and internal restructuring of the casein micelle and a more compact structure, resulting in less water associated with the protein (Moitzi et al., 2011). In samples C:P82 with ph 5.8, a higher water mobility and higher non-soluble whey protein level (Table 6) was observed due to 300 MPa UHPH treatment. The reverse effect of UHPH treatment was observed for samples C:P93 at ph 5.8. A generally higher solubilisation of proteins was observed for samples UHPH treated at ph 5.8 (Table 5). A significant different apparent viscosity was measured for samples at low ph 5.8 and UHPH treatment at 300 MPa and a higher apparent viscosity was measured in the samples with high casein content compared to samples with high whey protein content. The difference in apparent viscosity for these two samples could be explained with a

8 150 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) combination of water binding capacity of casein and whey proteins, solubilisation of proteins and denaturation of whey protein in sample C:P82. Our hypothesis is that a random protein structure is formed due to solubilisation, denaturation and aggregation of casein and whey proteins. The lower apparent viscosity in C:P82 was explained as, at low ph, a larger part of the whey protein will be associated with the casein and UHPH treatment results in solubilisation and denaturation. The whey protein can form complexes with κ-casein via disulphide bonds and also with other whey proteins (Hyslop, 2003) in the soluble phase. This indicates that, a more inhomogeneous re-association can take place following UHPH treatment, because of more complex protein solution containing various caseins and whey proteins. Changes in relaxation rates with whey protein denaturation have been reported (Lambelet et al., 1992), and the difference in water mobility for C:P82 at ph 5.8 with and without UHPH treatment could relate to denaturation of whey protein with UHPH treatment. In the system with low whey protein content (C:P93) a solubilisation of protein and aggregation into macro micelles could take place without any change in the water mobility, due to the low whey protein content. The higher apparent viscosity in samples C:P93 is due to solubilisation of caseins, inhomogeneous re-association due to solubilisation of caseins and water binding of caseins High field nuclear magnetic resonance The samples were analysed by 1 Hand 31 P NMR spectroscopy (Figs. 4 and 5), to achieve information about the changes in molecular structures as a result of various treatments. In the 1 H NMR spectra (Fig. 4) the most intense resonances originate from the partly suppressed water (~5.0 ppm) and lactose ( , 4.46, 4.69 and 5.24 ppm), whereas the less intense resonances shown in the inserts mainly originate from the proteins. The spectra of samples at ph 5.8 and ph 6.2 furthermore contain citric acid (resonances around 2.53 and 2.73 ppm). Besides that only minor ph related changes were observed for the resonances around 5.8 and 8.5 ppm. In the 31 P NMR spectra (Fig. 5) the most intense resonance originates from the serum phosphate. The chemical shift of this resonance is highly sensitive to changes in ph which is reflected by the change in chemical shift from 3.5 ppm at ph 5.8, 3.75 ppm at ph 6.2 and 4.5 ppm at ph 6.7. A resonance at 2.98 ppm tentatively assigned to glucose-1-phosphate was also present in all spectra. More interestingly, a range of resonances with a chemical shift 1 2 ppm higher than the serum phosphate was observed. These originate from serine-phosphate (SerP) (Belton et al., 1985; Ishii et al., 2001) and to compare the resonance of the serinephosphate in the different spectra, a relative integral between SerP and glucose-1-phosphate was calculated and shown in the spectra 1 H NMR Skim milk x25 x5 C:P93_6.7_300 C:P82_6.7_300 x25 x5 C:P93_6.2_150 C:P82_6.2_150 C:P93_5.8_300 C:P82_5.8_300 C:P93_6.7_0(2) C:P82_6.7_0(2) C:P93_6.7_0(1) C:P82_6.7_0(1) C:P93_5.8_0 C:P82_5.8_ ppm Spectrometer: 11.7 T, 6.0C ppm Fig. 4. Liquid-state 1 H NMR (11.7 T) spectra recorded at 6 C. Inserts are vertically scaled according to the numbers by the insert. Spectra of skim milk, samples with ph 5.8 without UHPH treatment (C:P93_5.8_0; C:P82_5.8_0), ph 6.2 with 150 MPa UHPH treatment (C:P93_6.2_150; C:P82_6.2_150), ph 5.2 with 300 MPa UHPH treatment (C:P93_5.8_300; C:P82_5.8_300), ph 6.7 with 300 MPa UHPH treatment (C:P93_6.7_300; C:P82_6.7_300), ph 6.7 without UHPH treatment (C:P93_6.7_0_1; C:P93_6.7_0_2; C:P82_6.7_0_1; C:P82_6.7_0_2).

9 H. Sørensen et al. / Innovative Food Science and Emerging Technologies 26 (2014) P NMR 151 Skim milk 63:37 C:P93_6.7_300 73:27 C:P82_6.7_300 83:17 C:P93_6.2_150 C:P93_5.8_300 80:20 85:15 C:P82_6.2_150 C:P82_5.8_300 89:11 88:12 C:P93_6.7_0(2) C:P93_6.7_0(1) 73:27 81:19 C:P82_6.7_0(2) C:P82_6.7_0(1) 74:26 70:30 C:P93_5.8_0 89:11 C:P82_5.8_0 91: ppm Spectrometer: 11.7 T, 6.0C ppm Fig. 5. Liquid-state 31 P (right column) NMR (11.7 T) spectra recorded at 6 C. Inserts are vertically scaled according to the numbers by the insert. Spectra of skim milk, samples with ph 5.8 without UHPH treatment (C:P93_5.8_0; C:P82_5.8_0), ph 6.2 with 150 MPa UHPH treatment (C:P93_6.2_150; C:P82_6.2_150), ph 5.2 with 300 MPa UHPH treatment (C:P93_5.8_300; C:P82_5.8_300), ph 6.7 with 300 MPa UHPH treatment (C:P93_6.7_300; C:P82_6.7_300), ph 6.7 without UHPH treatment (C:P93_6.7_0_1; C:P93_6.7_0_2; C:P82_6.7_0_1; C:P82_6.7_0_2). (Fig. 5). The SerP resonances for skim milk were of lower intensity compared especially to samples at ph 5.8 (C:P_5.8_0 and C: P_5.8_300). Furthermore a lower chemical shift was observed with decreased ph in accordance with Ishii et al. (2003). The reduction of the SerP content with decreasing ph observed in the 31 P NMR spectra is most likely related to the binding affinities of the phosphate binding sites in casein. Amongst the SerP resonances an intense resonance is observed in the C:P82 spectra compared to the C:P93 (Fig. 5, marked with a star). One of the major differences between the C:P82 and the C:P93 samples was that caseino-phospho-peptides (CPP) from the whey fraction was added to the C:P82 samples. Therefore the intense SerP resonance is tentatively assigned to CPP. It is known that with microfiltration, some soluble casein fragments will follow the whey protein in the filtration process (Kønigsfeldt, 2014). Phosphorous is bound to the casein via monoester linkages to seryl residues and stabilizes the amorphous calcium phosphate. The extent of phosphorylation is dependent on casein type and CPPs can arise from degradation of α s1 -, α s2 - and β-caseins. Overall the 1 Hand 31 P NMR spectra revealed only minor structural effectsofaddingccpandchangingph,butnoeffectofuhphtreatment, which indicate that UHPH treatment did not induce observable structural changes in the molecular structure of the individual proteins in the samples. 4. Conclusion The effects of whey protein content were studied in milk casein concentrate solution with two different ratios of casein-to-protein of, respectively, low 0.93 and high 0.82 and three different levels of ph and UHPH treatment. The chemical composition of the samples with two different levels of whey proteins was similar, but a higher casein and mineral level content was measured in the samples with high caseinto-protein ratio, At ph 5.8 and processing by UHPH treatment a higher solubilisation of proteins was observed in samples with high casein content and significantly more denaturation of whey protein was measured in samples with high whey protein. The highest apparent viscosity was detected in the samples with high casein at ph 5.8 with 300 MPa UHPH treatment, followed by samples with high whey protein at ph 5.8 with 300 MPa UHPH treatment, both exhibiting shear thinning. The other samples were characterised by similar apparent viscosity and exhibited Newtonian behaviour. This effect was explained by water binding of casein and solubilisation of proteins in the samples with high casein, while in the samples with high whey protein content the higher apparent viscosity was explained by denaturation of whey protein and solubilisation. Furthermore significantly larger particle sizes were observed for samples at ph 5.8 subjected to UHPH treatment. The water mobility determined by 1 H NMR relaxometry was strongly ph