Effect of dipotassium phosphate addition and heat on proteins and minerals in milk protein beverages

Our objective was to determine the effects of dipotas-sium phosphate (DKP) addition, heat treatments (no heat, high temperature, short time [HTST]: 72°C for 15 s, and direct steam injection UHT: 142°C for 2.3 s), and storage time on the soluble protein composition and mineral (P, Ca, K) concentration of the aqueous phase around casein micelles in 7.5% milk protein-based beverages made with liquid skim milk protein concentrate (MPC) and micellar casein concentrate (MCC). Milk protein concentrate was produced using a spiral wound polymeric membrane, and MCC was produced using a 0.1-µm ceramic membrane by filtration at 50°C. Two DKP concentrations were used (0% and 0.15% wt/wt) within each of the 3 heat treatments. All beverages had no other additives and ran through heat treatment without coagulation. Ultracentrifugation (2-h run at 4°C) supernatants of the beverages were collected at 1, 5, 8, 12, and 15-d storage at 4°C. Phosphorus, Ca, and K concentrations in the beverages and supernatants were measured using inductively coupled plasma spectrometry. Protein composition of supernatants was measured using Kjeldahl and sodium dodecyl sulfate-PAGE. Micellar casein concentrate and MPC beverages with 0.15% DKP had higher concentrations of supernatant protein, Ca, and P than beverages without DKP. Protein, Ca, and P concentrations were higher in MCC supernatant than in MPC supernatant when DKP was added, and these concentrations increased over storage time, especially when lower heat treatments (HTST or no heat treatment) had been applied. Dipotassium phosphate addition caused the dissociation of α S -, β, and κ-casein, and casein proteolysis products out of the casein micelles, and DKP addition explained over 70% of the increase in supernatant protein, P, and Ca concentrations. Dipotassium phosphate could be removed from 7.5% of protein beverages made with fresh liquid MCC and MPC (containing a residual lactose concentration of 0.6% to 0.7% and the proportional amount of soluble milk minerals), as these beverages maintain heat-processing stability without DKP addition.


INTRODUCTION
Protein claims have become an important factor in many consumers' choice of food and beverage products.According to a 2021 survey, 50% of consumers associate protein with a healthy diet (Kerry Group plc., 2021).International Food Information Council found that protein was the most sought-after nutrient by US consumers, with 59% of respondents in a 2020 survey reporting that they sought to consume more proteins (IFIC, 2022).Consumer desire for more protein has fueled the rising demand for ready-to-drink (RTD) highprotein beverages.The global RTD protein beverage market is projected to grow at a compounded annual growth rate of 7.7% between 2022 and 2027 (Mordor Intelligence, 2022).
Milk-based high-protein beverages can be produced from micellar casein concentrates (MCC) and milk protein concentrates (MPC).Both are milk-based protein ingredients that are produced using membrane filtration technologies.Microfiltration (MF) is used to produce MCC, which contains as much as 95% casein as a percentage of true protein (CN%TP) content (American Dairy Products Institute [ADPI], 2021).Ultrafiltration is used to produce MPC, which contains the same CN%TP content as that found in skim milk, i.e., about 80 to 82% (ADPI, 2021).The MCC and MPC may be manufactured as either fresh liquid concentrates or as powders.Fresh liquid concentrates are used directly or stored refrigerated, whereas powders Effect of dipotassium phosphate addition and heat on proteins and minerals in milk protein beverages Joice Pranata, 1 Hayden Hoyt, 2 MaryAnne Drake, 2 and David M. Barbano 1 * are shelf stable and can be produced by spray drying the concentrates following the filtration process.Although powders have longer shelf life and are lighter to transport than liquid concentrates, fresh liquid concentrates may have different properties than powders because they typically have fewer off flavors (Smith et al., 2016;Carter et al., 2018), require no rehydration step (Amelia and Barbano, 2013;McSweeney et al., 2022), and have better heat stability (Beliciu et al., 2012).
Commercial milk-based beverages formulated using MCC or MPC are typically heat treated to ensure safety from microbiological pathogens and to extend shelf life.The HTST thermal treatment can be used to produce pasteurized milk-based beverages with a shelf life of 2 weeks or longer when stored at 4°C (Fromm and Boor, 2004).Direct steam injection (DSI) can be used to produce ultrapasteurized beverages with a shelf life of 60 d or longer at 4°C or shelf-stable beverages with a shelf life of 6 mo or longer if packaged aseptically (Rysstad and Kolstad, 2006;Fernández García and Riera Rodríguez, 2014).In HTST pasteurization, the beverage must be heated at 72°C or higher for at least 15 s whereas in ultrapasteurization, the beverage must be heated at 138°C or higher for at least 2 s (FDA, 2022a).
Heat treatment may have detrimental effects on the stability of milk-based beverages.At high-protein content (≥5% casein), MCC beverages were reported to have poor heat stability, especially if the beverages were produced from reconstituted powders (Beliciu et al., 2012;Sauer and Moraru, 2012).Coagulation, formation of particle aggregates, and increase in viscosity were exacerbated in spray dried MCC compared with liquid MCC (Beliciu et al., 2012).These effects were suggested to occur due to a shift in mineral equilibrium as calcium phosphate solubility decreased at high temperatures and due to casein micelle dissociation (Beliciu et al., 2012;Sauer and Moraru, 2012).Heat coagulation has also been reported for both reconstituted MPC powder (Crowley et al., 2014) and liquid MPC (Tari et al., 2021).
To ensure heat-processing stability, calcium chelating salts (phosphates and citrates) are often added to milk-based beverage formulations (Augustin and Clarke, 1990;Fox et al., 2015).Addition of phosphate and citrate salts at low concentrations (less than 60 mEq/L) has been reported to improve the heat stability of beverages as demonstrated by an increase in heat coagulation time compared with beverages without salt addition (Meena et al., 2016;Renhe et al., 2018).The calcium chelating salts improved the heat stability of concentrates by reducing calcium ion activity and preventing protein aggregation (Augustin and Clarke, 1990).Calcium chelating salts have also been reported to enhance the dissolution and reconstitution of concentrate powders (Schuck et al., 2002;Ramchandran et al., 2017;Meena et al., 2018).Added calcium chelating salts may bind free soluble calcium, trigger the dissociation of colloidal calcium, form associations with caseins, and cause shifts in pH, but their exact interactions in milk systems and the relative contributions of each of the possible interactions are still not clearly understood (Kaliappan and Lucey, 2011).Removal of calcium by acidification with CO 2 injection has been shown to improve the heat stability and solubility of MCC and MPC powders (Marella et al., 2015;Sunkesula et al., 2021;Kommineni et al., 2022).The enhanced solubility properties of the powder could be attributed to the dissociation of soluble caseins upon removal of calcium (Marella et al., 2015).
Dipotassium phosphate (DKP) is one type of calcium chelating salt that has been used to improve heat stability of commercial milk-based beverages.Dipotassium phosphate has a generally recognized as safe status and is used extensively in shelf-stable dairy products (FDA, 2022b).Addition of calcium chelating salts such as DKP could be expected to influence the composition of soluble proteins and minerals in the continuous phase of the beverage (i.e., aqueous phase surrounding the casein micelles) by binding calcium and changing casein micelle structure.Changes in protein and mineral composition in the beverage continuous phase could cause perceptible changes to sensory and physicochemical properties that influence consumer acceptance of the beverage product.Supernatants collected from ultracentrifugation of these beverages can be used to represent the continuous phase surrounding the casein micelles.Higher protein concentration in the supernatant phase of liquid MCC has been associated with higher apparent viscosity in the liquid MCC material (Dunn et al., 2021;Pranata et al., 2023).
Previous studies evaluating effects of calcium chelating salt addition and heat treatment on casein micelle stability were mainly done using reconstituted MCC and MPC powders instead of fresh liquid concentrates (de Kort et al., 2011;Kaliappan and Lucey, 2011;McCarthy et al., 2017;Pandalaneni et al., 2018).Due to the high thermal stress and other modifications imposed on powder-based concentrates during the drying process, beverages made from reconstituted MCC and MPC powders might have poorer stability compared with beverages made from fresh liquid concentrates.While chelating salts have been reported to generally improve heat stability of beverages made using reconstituted powders, they could have different effects on beverages made from fresh liquid concentrates, which may already have sufficient heat stability without requiring salt addition.Addition of 0.15% DKP to MCC and MPC-based beverages made from liquid concentrates was found to increase the apparent (instrumental) and sensory viscosity of these beverages (Hoyt et al., 2023).
The necessity of the addition of chelating salts such as DKP to milk-based beverages produced from liquid concentrates should be carefully evaluated, and their usage should be avoided when possible, considering consumers' unfavorable perception of additives in food (Asioli et al., 2017).Understanding the influence of chelating salts on casein micelle stability and the partitioning of proteins and minerals in milk-based beverages following thermal treatment can aid processors in making decisions on their inclusion in beverage formulation.The objective of our study was to determine the effects of DKP addition, heat treatment (DSI, HTST, and no heat treatment), and storage time (at 4°C) on the protein and mineral composition in the aqueous phase surrounding the casein micelles of 7.5% protein beverages made with liquid MCC and MPC.

MATERIALS AND METHODS
No human or animal subjects were used, so this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.

Experimental Design
Liquid MCC and MPC beverages containing about 7.5% protein were produced from HTST pasteurized skim milk using a 3X, 3-stage ceramic MF process and a 3X, 3-stage polymeric UF membrane process, respectively (Hoyt et al., 2023).There were 6 treatments of MCC and MPC beverages, and a full factorial design was used to determine the effects of 0.15% DKP addition at 2 levels (added or not added) and heat treatment at 3 levels (no heat treatment, HTST, and DSI) according to the design described by Hoyt et al. (2023).Ultracentrifugation (UC) supernatants were collected for each of the MCC and MPC beverage treatments after 1, 5, 8, 12, and 15 d of storage at 4°C.The CP content of the supernatants was determined based on their Kjeldahl nitrogen.The composition of the proteins in the supernatants was determined by SDS-PAGE analysis.The concentration of phosphorus (P), calcium (Ca), and potassium (K) were determined by inductively coupled plasma (ICP) analysis.The entire experiment was replicated 3 times for MCC and 2 times for MPC beverages.

Production of Liquid Milk Protein Concentrate
Liquid MPC was produced by the UF of pasteurized skim milk using 10 kDa nominal molecular weight cutoff polyethersulfone polymeric membranes (TRISEP DS UP010 3883-31, Mycrodyn-Nadir US Inc.).Configuration, operation, and cleaning procedures were as described by Hoyt et al. (2023).

DKP Addition, Heat Treatment (HTST and DSI), and Storage Conditions
Liquid MCC and MPC were split into 6 treatment batches.In 3 of the batches, dipotassium phosphate was added to 0.15% concentration.Three different heat treatments (DSI [140°C for 2.3 s], HTST [72°C for 15 s], and no heat treatment) were applied toward each of the DKP added and no DKP batch pairs.Beverages were preserved with the addition of 1 mL of 10% wt/vol aqueous thimerosal (Thermo Fisher Scientific) per 1,000 g of beverage.The beverage samples were stored at 4°C and supernatants were collected from the beverages on d 1, 5, 8, 12, and 15 of storage for analysis.Dipotassium phosphate addition, heat processing, and storage conditions were performed as described by Hoyt et al. (2023).

UC of Liquid MCC and MPC to Collect Supernatants
Before centrifugation, the centrifuge rotor (T29-8x50 Super Speed Rotor, Thermo Fisher Scientific) and beverage samples were both tempered at 4°C for at least 12 h.The centrifuge (Sorvall Lynx 6000 Ultracentrifuge System, Thermo Fischer Scientific) was set to 4°C and allowed to temper for 1 h before use.Approximately 40 g (±1 g) of beverage was weighed into each centrifuge tube (Nalgene Oak Ridge PPCO 3139-0050 Tube, Thermo Fischer Scientific).Centrifuge tubes containing samples were paired based on weight (0.02 g difference max) to balance the centrifuge rotor (4 pairs of tubes total).Tube pairs were placed in the rotor opposite one another and were centrifuged at 100,605 × g for 2 h at 4°C.The centrifuge tubes were removed from the rotor, and the top 15 mL of supernatant was collected carefully by removal from the top liquid surface from each tube in three 5-mL portions with a 5-mL pipettor for further chemical analysis.This process was repeated 3 more times (4 total vials of test material) so that there were 2 vials each containing 30 mL of supernatant collected for each treatment for further analysis.This was done for all treatments.The entire process from production to UC supernatants collection was repeated 3 times for MCC beverages and 2 times for MPC beverages.Six treatment samples of supernatants across 5 time points were collected for a total of 30 samples across each replicate for each of the MCC and MPC beverages.The supernatants were used to represent the aqueous phase around the casein micelles.

Chemical Analysis of Supernatants
Protein and SDS-PAGE Analysis.MCC and MPC supernatants were analyzed in duplicate by Kjeldahl total nitrogen (TN; AOACI, 2019, method number 990.20).The CP was calculated as TN multiplied by 6.38.Supernatants were analyzed using SDS-PAGE to determine the composition of individual protein types including α S -CN, β-CN, κ-CN, casein proteolysis products (CNPP), and serum proteins (SP) in the supernatant protein.SDS-PAGE analysis was performed using the procedure of Verdi et al. (1987) applied to MCC and MPC supernatant samples with adjustments on sample protein loading according to Carter et al. (2021).After fixing, staining, and destaining, the electrophoresis gels were scanned with USB GS 800 Densitometer using Quantity 1 1-D Analysis software (Bio-Rad Laboratories Inc.) and analyzed quantitatively to obtain the relative proportions of individual protein types in the sample as described by Carter et al. (2021).The relative proportions of each protein type in the supernatant were multiplied by the CP content of the supernatant that was determined by Kjeldahl analysis to obtain the absolute concentration of each protein type in the supernatant.
Mineral Analysis and pH.Phosphorus, calcium, and potassium concentration in the MCC and MPC beverage supernatants were measured in duplicate using ICP analysis.Analysis was performed using the methods described by Hoyt et al. (2023).The pH of the supernatants was measured at room temperature (22°C) using an Accumet pHmeter (915, Fisher Scientific) with a combination electrode (HA405-DXK-S8/120, Mettler Toledo).The pH meter was calibrated using a 2-point calibration with pH 4 and pH 7 buffer solutions (SB101 and SB107 at 22°C, Fisher Scientific) as references.
Freezing Point.The freezing points of the supernatants were measured using an Advanced cryoscope (Model 4250, Advanced Instruments Inc.).For each measurement, 2.5 mL of supernatants were used.The cryoscope was calibrated using 422 and 621 reference solutions (3LA023 and 3LA033, Advanced Instruments Inc.), and the instrument performance was checked using Lactrol 530 reference solution (3LA030, Advanced Instruments Inc.).

Statistical Analysis
Data were analyzed using SAS version 9.4 (SAS Institute Inc.).Data for MPC-based beverages and MCC-based beverages were analyzed separately.Proc GLM of SAS was used to determine the effects of independent variables heat treatment, DKP addition, replicate, and storage time on dependent variables CP, α S -CN, β-CN, κ-CN, casein proteolysis products, serum protein, phosphorus, calcium, and potassium concentration in the supernatants.Heat treatment (no heat, HTST, DSI), addition of DKP (none, 0.15% added), and replicate were included in the model as categorical variables with all 2-way interaction terms.Refrigerated storage time (0 to 15 d) was included as a continuous variable in the model and was mean centered with a mean value of zero to minimize effects of co-linearity (Glantz and Slinker, 2001).Both the linear and quadratic terms for mean-centered time, and all 2-and 3-way interactions were included in the model.If the F-value for the full model was significant (P < 0.05), then significance (P < 0.05) of each independent variable and its interactions were determined.The effects of the categorical variables and their interactions were tested for significance using heat treatment × DKP × replicate as the error term, whereas the effect of time (linear and quadratic) as well as its interactions with categorical variables were tested for significance using the full model error.A backward stepwise process was done to remove all nonsignificant terms from the model to produce a final reduced model with type III sum of squares (SS) table and least squares means table for each dependent variable analyzed.Relative percentage type III SS for each dependent variable analyzed were calculated by dividing the type III SS for each significant term in the final reduced model with the total type III SS and multiplying them by 100.Relative percentage type III SS shows the relative amount of variation explained by each term in the model.

CP of UC Supernatants
Ultracentrifugation of the protein beverages separates the casein micelles (i.e., dispersed phase) from the supernatants (i.e., continuous phase).The beverage supernatants reflect the aqueous phase surrounding the casein micelles.The supernatant CP concentration represents the amount of serum protein and dissociated caseins in the beverage continuous phase and is shown in Table 1.
Addition of DKP increased (P < 0.05) the supernatant protein concentration in MCC and MPC beverages (Tables 1, 2, and 3; Figures 1 and 2).Dipotassium phosphate accounted for 81% and 77% of the variation in MCC and MPC supernatant CP concentration, respectively (Tables 2 and 3).The addition of DKP to MCC and MPC beverages caused proteins to dissociate out of the casein micelles.Dipotassium phosphate could act as a calcium chelating salt and cause casein monomers to dissociate out of the micelles by shifting the beverage mineral equilibrium.The role of calcium chelating salts in the dissociation of casein micelles in milk systems have been reported previously (Kaliappan and Lucey, 2011;Culler et al., 2017).
The removal of minerals and the higher protein content in MCC and MPC beverages may have also disrupted the calcium phosphate equilibrium of the micelles and contributed to the observed strong effect of DKP in the beverages (Tables 2 and 3).As a comparison, the effect of DKP was negligible in skim milk, in which the electrostatic environment is richer and the protein concentration is lower compared with the concentrates.The protein content of the skim milk ranged between 3.3% and 3.6%, and the supernatant Numbers not sharing a common superscript within a column (separately within effects of heat treatment and DKP addition) are different (P < 0.05).CP contents of skim milk with and without DKP were approximately the same at 1.1 to 1.3% from d 1 to d 15 of cold storage.In contrast, the mean protein contents of MCC and MPC supernatants were about 3% and 1% higher, respectvely, with DKP addition compared with without DKP addition (Table 1).Calcium phosphate equilibrium influences casein micelle stability.Addition of salts such as DKP and removal of minerals through processes such as microfiltration and UF disrupts this balance and can lead to protein dissociation out of the micelles.Pranata et al. ( 2023) reported higher protein dissociation during cold gelation in MCC beverages compared with skim milk due to higher protein content and reduced ionic strength of the MCC beverages.
To the authors' knowledge, a comparison between the dissociation of proteins between MCC and MPC beverages with and without chelating salts has not been previously reported.Crude protein % in the supernatant  was about the same in MCC and MPC without added DKP, and a portion (23% in MCC and 57% in MPC) of the protein in the supernatant was milk SP (Figures 1 and 2).Crude protein concentration was higher in MCC supernatant than MPC supernatant with added DKP, indicating that more protein moved from the micelles into the beverage continuous phase in MCC than in MPC (Figures 1 and 2).The higher protein content in the MCC supernatant compared with MPC supernatant with DKP added may be partly due to the higher starting casein content of the MCC beverages with the MPC beverages.The remaining difference in MCC supernatant protein content may be explained by the higher casein proteolysis products in MCC beverage.Casein proteolysis products made up most the MCC supernatant proteins with DKP added (Table 1).On average, with DKP addition, 33% of MCC supernatant proteins and only 11% of MPC supernatant proteins were casein proteolysis products (Table 1).
Protein concentration in MCC and MPC supernatant increased (P < 0.05) with time, especially with DKP and low heat treatment in MCC beverages (Tables 2  and 3, Figures 1 and 2).With DKP addition, beverages treated by HTST or without any heat treatment showed more protein dissociation over time compared with beverages treated by DSI (Figures 1 and 2).The DSI treatment may have resulted in modifications to the surface of the casein micelles that prevented continued protein dissociation over time.Upon heating above ca.70°C, whey proteins denature and can modify the surface of casein micelles by forming disulfide bonds with κ-CN (Dalgleish and Corredig, 2012).The DSI treated beverages started with a higher level of dissociated proteins in the supernatant, but a heat induced coating of the casein micelles with whey proteins may have prevented further protein release over time (Figure 1 and 2).
α S -, β-, and κ-CN.In the presence of DKP without any heat treatment, at d 1 of storage, α S -CN accounted for 20% and 8% of the supernatant protein in MCC and MPC, respectively, and only 4 to 5% of the supernatant protein in skim milk.Compared with skim milk, higher levels of α S -CN were observed in the supernatant of MCC and MPC, especially with DKP addition.In skim milk with and without DKP, β-CN comprised 16 to 20% of total supernatant proteins and was the primary casein found in the supernatant, consistent with previous reports (Davies and Law, 1983).At d 1 of storage, β-CN accounted for 16% and 18% of the supernatant protein in unheated and DKP added MCC and MPC, respectively.The composition of β-CN in the beverage supernatant appeared to be maintained as in the original skim milk while supernatant α S -CN increased in the beverage.This may be due to the higher sensitivity of α S -CN to alterations in ionic strength and mineral equilibrium during filtration and DKP addition.α S1and α S2 -CN have more clusters of phosphoseryl residues than β-CN while κ-CN has none (Swaisgood, 1993), so α S -CN may be more sensitive to alterations in calcium concentration and more susceptible to dissociation with the addition of calcium chelating salts such as DKP.
The concentration of α S -, β-, and κ-CN in MCC supernatant was higher than in MPC supernatant (Table 1).Dipotassium phosphate accounted for 90, 83, and 78% of the variation in α S -, β-, and κ-CN concentrations in MCC supernatant, respectively (Table 2).Dipotassium phosphate accounted for 60, 68, and 28% of the variation in MPC supernatant α S -, β-, and κ-CN concentrations, respectively (Table 3).Addition of DKP increased the concentration of α S -, β-, and κ-CN in MCC and MPC supernatants (P < 0.05) relative to no DKP addition (Table 1).The effect of DKP on α S -CN concentration in the supernatant was larger in MCC than in MPC (Table 1).Concentration of α S -, β-, and κ-CN increased (P < 0.05) with time for both MCC and MPC (Tables 2 and 3), especially with DKP which was similar to the increase in protein concentration in the MCC and MPC supernatants (Figures 1  and 2).
Concentration of α S -, β-, and κ-CN in MPC supernatant increased (P < 0.05) with higher heat treatment (Tables 1 and 3).Supernatant κ-CN concentration increased with higher levels of heat treatment in MPC (Table 1), while there was less effect of heat on MCC, and there was no heat × DKP interaction in MCC (Table 2).In MPC, there was a strong effect of heat (52% of variation) and a heat × DKP interaction (9% of variation; Table 3).Unlike in MCC, higher heat treatment may have caused more heat-denatured serum protein to form disulfide bonds with κ-CN in MPC.
Dissociation of caseins out of micelles may have implications on the physical and sensory properties of the beverages, namely viscosity and appearance.Hoyt et al. (2023) reported that apparent (instrumental) and sensory viscosity increased with DKP addition and storage time of MCC and MPC beverages.Casein micelle dissociation with DKP addition may contribute to viscosity increase through the formation of casein monomer interactions outside of the micelles, and the formation of loose entanglements of caseins outside of the micelles.
Above pH 6.8, DKP exhibits 2 negative phosphate charges, and it could form calcium-CN phosphate cross-links in the continuous phase that could result in increased beverage viscosity.Formation of calcium-CN phosphate complexes by polyphosphates has been shown previously in reconstituted MCC (de Kort et al., 2011) and MPC (Pandalaneni et al., 2018) powders and was associated with an observed increase in apparent viscosity.Addition of calcium chelating salts has also been shown to increase α S -, β-, and κ-CN concentration in the supernatant of reconstituted skim milk (Sikand et al., 2016) andMPC (McCarthy et al., 2017) powders, which was accompanied by an increase in apparent viscosity (McCarthy et al., 2017).
Studies using transmission electron microscopy in skim milk acidified to pH 5.2 (McMahon et al., 2009) and UF skim milk diafiltered at pH 5.4 to 5.8 (Liu et al., 2017) have shown that casein micelle dissociation is accompanied by the formation of loose entanglements of the proteins that were released.Proteins released from casein micelles due to DKP addition could also form loose entanglements in the beverage continuous phase and lead to an increase viscosity.
The effect of casein micelle dissociation on viscosity is also influenced by protein concentration.At higher protein concentrations, the dispersed phase volume in the beverage is also larger and casein micelles are packed more tightly in the beverage (Bouchoux et al., 2009).Closer packing of the casein micelles in the beverage increases the likelihood of protein entanglement and cross linking outside the micelles.Using an estimated voluminosity of 4.4 mL/g (de Kruif, 1998), the dispersed casein micelle phase volume of 7.5% protein beverage is around 0.33.At phase volumes of 0.25 to 0.54, casein micelle suspensions exhibit concentration dependent viscosity similar to that of hard sphere suspensions (Dahbi et al., 2010;Corredig et al., 2019).
Dissociation of caseins out of micelles may also have implications on the appearance of beverages.In a separate analysis, sensory opacity and whiteness (Hunter L value) of MCC and MPC beverages decreased with DKP addition and storage time (Hoyt et al., 2023).As caseins and calcium phosphate dissociate out of the casein micelles, the micelles become more translucent and light scattering is reduced.Studies in reconstituted MPC (McCarthy et al., 2017), reconstituted MCC (de Kort et al., 2011), and diluted skim milk (Culler et al., 2017) have also demonstrated that casein micelles dissociate with addition of chelating salts and this was accompanied by a decrease in turbidity (whiteness).Culler et al. (2017) reported that DKP addition caused an over 30% decrease in turbidity of diluted milk solution.McCarthy et al. (2017) found that sodium hexametaphosphate and trisodium citrate addition led to a decrease in turbidity and L* value in reconstituted MPC.
Noncasein Protein.With increasing levels of heat treatment, the noncasein protein concentration (NCP) in both the MCC and MPC supernatants decreased (P < 0.05).Without heat, approximately 18 and 57% of the proteins in the MCC and MPC supernatants were noncasein protein whereas with DSI treatment, only 15 and 41% of the proteins in MCC and MPC supernatants were NCP, respectively (Table 1).
The NCP is composed mainly of whey proteins.It has been well-documented that whey proteins can form disulfide bonds with κ-CN upon heat treatment (Purkayasthar et al., 1967).In both MCC and MPC beverages, as heat treatment increased, more of the whey protein denatured and was covalently bound to the casein micelles and was collected in the ultracentrifugation pellet (casein micelle phase) instead of the supernatant (continuous aqueous phase).Similar to this observation, Lin et al. (2018) found a reduction in serum protein concentration in the supernatant of reconstituted MPC powder with increasing heat treatment (from 72°C to 85°C) that was attributed to increased whey protein denaturation.

Casein Proteolysis Products
Casein proteolysis product concentration was about 5 times higher in MCC compared with MPC in supernatants (Table 1, Figures 3 and 4).The CNPP concentration in the MCC supernatant with added DKP increased up to about 2.5% after 15 d storage.Dipotassium phosphate accounted for 68% and 43% of the variation in MCC and MPC supernatant CNPP concentration, respectively (Tables 2 and 3).Addition of DKP increased the concentration of casein proteolysis products in MCC (P < 0.05) and MPC (P < 0.05) supernatants (Tables 2 and 3), but the magnitude of the increase was much larger for MCC than MPC (Table 1).The CNPP also increased over storage time (P < 0.05) for both MCC and MPC supernatants (Tables 2 and 3), but the increase in concentration was much larger for MCC (Figure 3) than MPC (Figure 4).Time accounted for 14% of the variation in CNPP supernatant concentration.Increase in CNPP percentage over storage time was higher in MCC (Figure 3) with DKP than MCC without DKP (demonstrating the Time × DKP interaction effect accounting for 5% of the variation in CNPP concentration).
Proteolysis in the beverage could be attributed to native plasmin activity and not microbial activity because the microbial load of the beverages was monitored to ensure contamination did not occur and a preservative (thimerosal) was added to the beverages to prevent microbial growth (Hoyt et al., 2023).The high levels of CNPP observed in the beverages, especially in MCC, may be attributed to enhanced plasmin activity during processing at the filtration temperature of 50°C, increased access of plasmin to casein substrates as casein micelles were dissociated, and the reduction of β-lactoglobulin that can act as plasmin inhibitor (Saint Denis et al., 2001).
Despite the high concentration of casein proteolysis products measured in the supernatants, a separate sensory analysis showed that bitterness was not detected in the beverages (Hoyt et al., 2023).This result suggests that even though there is a high concentration of CNPP measured (i.e., high extent of proteolysis), the casein breakdown has not yet progressed far enough to form the small peptides that elicit sensory defects.Supernatant CNPP measured using SDS-PAGE ranged in molecular weight from ca. 7 to 23 kDa with most of the CNPP in the 11-12 kDa size range and referred to as "CNPP 6."These CNPP likely originated mainly from the plasmin proteolysis of β-CN that produced larger peptide fractions of Γ 1 -, Γ 2 -, and Γ 3 -CN in addition to proteolysis of α S1 -CN that produced peptide fractions of 10 kDa to over 20 kDa in sizes (Verdi et al., 1987).
Mineral (P, Ca, K).Total calcium in the MCC and MPC beverages were approximately 2,400 and 2,100 mg/L, respectively, and total phosphorus in the MCC and MPC beverages with 0.15% DKP were approximately 1,800 and 1,500 mg/L, respectively (Hoyt et al., 2023).With DKP, approximately 50% and 30% of total Ca and 55% and 40% of total P in MCC and MPC beverages, respectively, were in the continuous phase, i.e., supernatant (data not shown).
Addition of DKP had a large impact on concentration of Ca in the supernatants and explained about 70% of the variation for both MCC and MPC (Table 4).There was a heat by DKP interaction (P < 0.05) for Ca concentration in MPC supernatants, and Ca concentration in both MCC and MPC supernatants increased (P < 0.05) with time (Table 4).
Phosphorus content of the supernatants was higher when DKP was added (Table 5), but the magnitude of increase in P concentration in the supernatants was larger than would be expected from the addition of DKP alone and would indicate that some of the colloidal phosphate in casein micelles was shifted into the aqueous phase around the micelles.Addition of 0.15% DKP in the beverage should increase the phosphorus content in the beverage by only 267 mg/L, but the phosphorus content in the supernatant with DKP increased more than this compared with supernatant without DKP.As expected, the concentration of K in both MCC and MPC supernatants was increased by addition of DKP (Table 5).
Phosphorus and calcium supernatant concentrations were higher (P < 0.05) with DKP added (Table 5).Dipotassium phosphate addition explained most the  Phosphorus and calcium concentration in the supernatants increased (P < 0.05) with time (Table 4), especially with DKP and lower heat (Figures 5-8).These effects were especially prominent in MCC supernatant (Figures 5 and 6).MCC supernatants also contained higher concentrations of P and Ca than MPC supernatants (Table 5).Overall, the trends in P and Ca concentrations mirror the effects of DKP, storage time, and heat observed in protein and casein concentrations of the supernatant.Chelation of ionic Ca in the continuous phase by chelating salts such as DKP causes a decrease in soluble calcium phosphate saturation and triggers release of colloidal calcium phosphate (Broyard and Gaucheron, 2015).Release of colloidal calcium phosphate disrupts the casein micelle structure and is associated with a decrease in whiteness due to the release of casein monomers (Mizuno and Lucey, 2005;Broyard and Gaucheron, 2015).Li and Corredig (2014) showed that removal of ionic calcium during UF and distilled water diafiltration of skim milk at its native pH resulted in release of colloidal calcium phosphate as the casein micelle structure was destabilized.Ferrer et al. ( 2011) also demonstrated a loss of colloidal Ca in skim milk following UF at native milk pH due to removal of ionic calcium.Similarly, we found that the chelation of the ionic calcium by DKP in milk beverages caused a shift in mineral equilibrium and resulted in the loss of colloidal Ca.Milk-based calcium can be separated to colloidal calcium (in the casein micelle phase) and soluble calcium (in the continuous phase) that usually exist in a Numbers not sharing a common superscript within each treatment category (heat treatment and DKP addition) within a column are different (P < 0.05).
1 Heat treatment = no heat, HTST, direct steam injection (DSI).  2 to 1 (colloidal to soluble) ratio in milk (Holt, 1981).Colloidal calcium exists mainly as calcium phosphate nanoclusters in the casein micelles, and soluble calcium exists as free calcium ions (Ca 2+ ) and complexes with citrate, phosphates, and phosphate esters (Boulet and Marier, 1960;Holt, 1981).Milk-based phosphorus can also be distinguished as colloidal and soluble phosphorus.Colloidal phosphorus usually exists as micellar calcium phosphate nanoclusters and as part of the casein phosphoserine residue, while soluble phosphorus may be present as free phosphate ions (HPO 4 2− and H 2 PO 4 − ) or calcium phosphate salts (CaHPO 4 ; Holt, 1992;Gaucher et al., 2007;Mekmene et al., 2009).
The mineral analysis of the supernatants in the current study does not distinguish between free ions, salts, and protein bound minerals of Ca and P. Ca and P may have been released with casein micelle dissociation as part of casein bound minerals, chelated complexes with DKP, or as free ions.The increasing casein content observed in the supernatant with DKP addition and storage time suggests that at least some of the Ca and P measured in the supernatants originated from phosphoserine residues of caseins and were protein bound Ca and P. In 5% reconstituted MPC solutions, Kaliappan and Lucey (2011) found that addition of 0.1% to 0.7% calcium chelating salts (binary  mixtures of sodium hexametaphosphate, disodium phosphate, and tetrasodium pyrophosphate) increased casein bound Ca and P concentrations in the MPC solution compared with without salt addition.Similarly, Mizuno and Lucey (2005) found that addition of 0.1% to 2.0% tetrasodium pyrophosphate and sodium hexametaphosphate caused an increase in the concentrations of casein bound Ca in 5% reconstituted MPC solutions.
pH.Both MCC and MPC supernatants had similar pH ranging from 6.9 to 7.1 (Table 5).There were no practical differences in pH across time of storage and across treatments as differences in pH between individual samples were less than 0.2 (data not shown).We concluded that pH did not contribute greatly to the changes in mineral equilibrium between micelles and the aqueous phase around casein micelles that occurred during time of beverage storage.While Ca and P content increased with time in the supernatants, the pH remained constant.Observed dissociation of protein, Ca, and P was not driven by changes in pH.

Freezing Point
Addition of DKP explained 99% and 95% of the variation in MCC and MPC supernatants freezing points, respectively (Table 4).The freezing points for supernatants with DKP were lower (i.e., more negative) due to the freezing point depression from the added salt.The mineral content of supernatants increased (P < 0.05) with time (Table 4) of storage (Figures 5-8) while the freezing point of the supernatants was constant across time and heat treatment as well as between MCC and MPC when considered within each category of DKP addition (0% and 0.15% DKP; Figure 9).As such, most of the minerals that dissociated into the supernatants during storage were not present as free ions, but instead were in the form of complexed minerals or bound to the proteins.Future studies are needed to investigate the effect of serum protein, casein proteolysis products, beverage mineral, soluble mineral, and lactose concentration on casein micelle dissociation in these high-protein beverages.

CONCLUSIONS
MCC and MPC beverages with 0.15% DKP had higher (P < 0.05) concentrations of supernatant protein, Ca, and P than beverages without DKP.Protein, Ca, and P concentrations were higher in MCC supernatant than in MPC supernatant when DKP was added, and these concentrations increased (P < 0.05) over storage time especially when lower heat treatments (HTST or no heat treatment) had been applied.Dipotassium phosphate addition caused the dissociation of α S -, β-, κ-CN, and casein proteolysis products out of the casein micelles, and DKP addition explained over 70% of the increase in supernatant protein, P, and Ca concentrations.Beverages (7.5% protein) were process stable when DKP was not added, so DKP could be removed from protein beverages made with fresh liquid MCC and MPC, containing a residual lactose concentration of 0.6 to 0.7% and the proportional amount of soluble milk minerals, to create beverages that have fewer additives and achieve a clean label to address consumer preference for clean label products.

Figure 1 .
Figure 1.Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on CP concentration (g/100 g supernatant) of micellar casein concentrate (MCC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.
Figure 2. Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on CP concentration (g/100 g supernatant) of milk protein concentrate (MPC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.
Figure 3.Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on casein proteolysis product concentration (g/100 g supernatant) of micellar casein concentrate (MCC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.
Figure 4. Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on casein proteolysis product concentration (g/100 g supernatant) of milk protein concentrate (MPC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.

Figure 5 .
Figure 5.Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on phosphorus (P) concentration of micellar casein concentrate (MCC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 3.
Figure 6.Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on phosphorus (P) concentration of milk protein concentrate (MPC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.

Figure 7 .
Figure 7. Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on calcium (Ca) concentration of micellar casein concentrate (MCC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 3.
Figure 8.Effect of heat treatment, added dipotassium phosphate (DKP), and days of storage (at 4°C) on calcium (Ca) concentration of milk protein concentrate (MPC) supernatant with and without added DKP and under 3 different heat treatments (HTST, no heat, and direct steam injection [DSI]); n = 2.
Pranata et al.: EFFECT OF DIPOTASSIUM PHOSPHATE Pranata et al.: EFFECT OF DIPOTASSIUM PHOSPHATE

Table 1 .
Pranata et al.:EFFECT OF DIPOTASSIUM PHOSPHATE Least squares means for supernatant CP and individual protein concentrations (g/100 g supernatant) for the effect of heat treatment and dipotassium phosphate for 7.5% protein beverage made with micellar casein concentrate (MCC) or milk protein concentrate (MPC) across 15 d of storage at 4°C

Table 2 .
Micellar casein concentrate (MCC) supernatants: individual protein concentrations (α S -CN, β-CN, κ-CN, casein proteolysis products, and serum protein) relative percentage of type III sum of squares (for factors with P < 0.05) explained by model factors

Table 4 .
Mineral content and freezing point relative percentage of type III sum of squares (for factors with P < 0.05) explained by model factors 1 for micellar casein concentrate (MCC) and milk protein concentrate (MPC) supernatants

Table 5 .
Pranata et al.:EFFECT OF DIPOTASSIUM PHOSPHATE Least squares means for supernatant mineral concentration, freezing point, and pH (at room temperature) for the effect of heat treatment 1 and dipotassium phosphate 2 for 7.5% protein beverage made with micellar casein concentrate (MCC) or milk protein concentrate (MPC) across 15 d of storage at 4°C