Novel details on the dissociation of casein micelle suspensions as a function of pH and temperature

Membrane filtration is a widespread process for fractionation and recombination of milk components. Although the dissociation of micellar caseins has been studied in detail in skim milk, it is important to better understand the dissociation dynamics occurring between the colloidal and noncolloidal fractions in systems of modified composition. This research aimed at understanding the dissociation of casein proteins in micellar fractions depleted of whey proteins. Casein micelle dispersions were tested at neutral pH and pH 6 (using glucono-δ-lactone as acidulant), after incubation at 4°C or 22°C, and compared with skim milk. The ionic composition of the serum phase was measured using inductively coupled plasma-mass spectrometry, and the protein distribution analyzed using reversed phase-HPLC coupled with mass spectrometry. When incubated at 22°C, there were no differences in casein micelle dissociation between skim milk and whey protein-depleted micelles (~2.6% dissociated casein). No additional dissociation occurred by lowering the pH from 6.8 to 6 at 22°C, albeit there were more soluble ions at low pH (71% Ca and 65% P). At 4°C, there was an increased amount of β-casein found in the serum phase (23–33% of total β-casein). In addition, there was an uneven dissociation behavior of the various genetic β-casein variants, whereof A2 was more readily released with cooling. In skim milk, approximately 22%, 18%, and 14% of κ, α S2 , and α S1 -caseins, respectively, were dissociated from the micellar phase upon cooling and acidification to pH 6.0. This was in contrast to whey protein-depleted casein suspensions, in which only 6%, 5%, and 3% of κ, α S2 , and α S1 -caseins, respectively, had dissociated. The results suggested that the whey proteins in the serum phase play a role in the equilibrium between colloidal and soluble caseins in milk. This is of great relevance in processes such as cold membrane fractionation, where more attention should be given to the protein composition in the serum phase, especially when concentration is combined with fractionation of the serum proteins


INTRODUCTION
The bovine casein micelle is composed of a heterogeneous network of casein proteins, water and salts.There are 4 main caseins: κ-, α S1 -, α S2 -and β-CN, whereof α S1 -and β-CN are the most abundant with nearly 4-fold the concentration relative to κ-and α S2 -CN (Fox et al., 2015).They are organized in colloidal particles, called casein micelles.Their precise 3-dimensional structure is still under debate.κ-Casein, mostly present on the surface of these protein particles, plays an important role in imparting their steric and electrostatic stabilization (Dalgleish and Corredig, 2012).In addition to caseins (about 80% of the total protein), skim milk also contains whey proteins (WP), which are globular proteins present in the soluble phase, under native conditions.The predominant WP are β-LG and α-LA, which together constitute approximately 20% of the total protein in bovine milk.Some of the most abundant minerals in milk are Ca (~1,200 mg/kg, 30 mmol/kg) and inorganic PO 4 (~2,000 mg/kg, 20 mmol/kg; Gaucheron, 2005).The latter refers to the PO 4 , which is not esterified to proteins.Approximately two-thirds of the total Ca ions and half of the PO 4 are present in the form of nanoclusters associated with the caseins via their phosphoserine groups (Davies and White, 1960;Holt, 1997).β-, α S2 -and α S1 -casein are the most phosphorylated caseins and thus mostly responsible for these ionic interactions.The remaining Ca and PO 4 ions are found in the serum phase, where Ca is mainly bound to citrate (~8.2 mM), but also complexed with PO 4 , chloride, α-LA, or it can simply be in its free ionic form (Ca 2+ ) (Barone et al., 2021).
In addition to Ca and PO 4 , milk contains also a variety of other salts and minerals, including monovalent cations such as Na, K, Mg, Fe, Cl, as well as organic salts such as citrate (Gaucheron, 2005).
Casein micelles can be promptly destabilized by, for example, a decrease in charge with pH, chelating agents such as EDTA, or by specific hydrolysis with chymosin (Dalgleish and Corredig, 2012).Changes in environmental factors such as storage temperature, pH, and salt composition of the soluble phase can cause subtle to severe changes to the equilibrium between the colloidal and soluble casein protein distribution.Storage at temperatures <10°C are known to increase the fraction of soluble β-CN, due to a weakening of hydrophobic interactions (Creamer et al., 1977).Cooling also causes diffusion of colloidal Ca and PO 4 causing a 10% increase of serum phase Ca and PO 4 (Gaucheron, 2005).Likewise, pH decrease also increases Ca and PO 4 solubility (van Hooydonk et al., 1986;Dalgleish and Law, 1989;Gaucheron, 2005).At pH 6.8 the soluble Ca concentration is about 200 mg/L, which increases to approximately 250 mg/L at pH 6.0 (Li and Corredig, 2014).The pH-dependent dissociation of Ca and PO 4 happens irrespective of temperature, in the range between 4 and 30°C (Dalgleish and Law, 1989).This gradual decrease continues until pH below 5.0 where most of the Ca is now dissociated from the casein micelles and can be found in the serum phase.Unlike ions, which show little dependence of temperature (Dalgleish and Law, 1989), the extent of casein dissociation with pH is dependent on temperature; at pH 5.5, whereas at 30°C, virtually no casein is released from the micellar phase, and at 4°C up to 60% of the casein is found in the soluble phase (Dalgleish and Law, 1988).Whey proteins associate with ions in the soluble phase (Rahimi-Yazdi et al., 2010).With heating (>70°C), β-LG is known to modify its structure causing the exposure of hydrophobic sites and reactive thiol groups that can aggregate with other proteins, through thiol-disulfide bridges with other thiol groups (e.g., on κ-CN, α-LA, or β-LG itself, or via other noncovalent interactions; Corredig and Dalgleish, 1996).Calcium-chelating agents have shown to cause casein dissociation (Marchin et al., 2007).It has also been demonstrated that addition of Ca or Na causes a pH decrease (Gaucheron, 2005;Barone et al., 2021).
The effect of protein and nanocluster dissociation in casein micelles in the absence of WP has yet to be studied in detail.It is generally thought that casein micelle dissociation is not affected by the absence or presence of WP, but only by pH and temperature.This is of particular significance in processes where via the use of large pore size membranes (microfiltration) it is possible to separate the WP while maintaining the native structure of the casein micelles (Mistry and Maubois, 1993).
Objective of this study was to address the effect of the presence of WP on casein dissociation dynamics as a function of temperature or pH.Casein suspensions were studied at their native pH, 6.8, and at a lower pH, 6.0, well above the caseins' isoelectric point.The results will contribute to a better understanding of the processing properties of casein micelle dispersions depleted of WP, such as micellar casein concentrates or isolates.

MATERIALS AND METHODS
Because no human or animal subjects were used, this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.
Low-pasteurized skim milk (72°C, 15 s) was obtained from the Arla Food Innovation Center (Agro Food Park, Aarhus, Denmark).Sodium azide was added at a concentration of 0.2 mg/mL for preservation.Ultrafiltration was conducted using a tangential flow filtration system (Vibro LE, SANI Membranes ApS, Allerød, Denmark) with hydrophilic polyethersulfone membranes (nominal molecular weight cutoff 30 kDa; membrane area of 0.35 m 2 ).Ultrafiltration was performed at 10°C with a 1-Pa transmembrane pressure (Coşkun et al., 2022) The permeate contained no WP, and an amount of free ions comparable to that of the original milk serum.
Casein micelle suspensions were prepared from skim milk by separating them from their serum phase by ultracentrifugation (at 100,000 × g for 1 h).Centrifugations were carried out at 22°C to avoid temperature dependent dissociation.The serum phase (centrifugal supernatant) was separated and substituted with UF permeate.These suspensions were then compared with control skim milk, still containing the original WP.The samples were then kept (on a rotary mixer, Fisher Scientific, Thermo Fisher) overnight either at 22°C or 4°C and incubated with (or without) 0.35% wt/wt of glucono-δ-lactone to reach pH 6.0.After overnight incubation the suspensions (± WP) were again separated by ultracentrifugation (as above).In this case, the centrifugation was carried out either at 22°C or 4°C.
The full suspensions and their respective centrifugal supernatants were analyzed for protein composition using reverse-phase (RP)-HPLC (Agilent Technologies) combined with a UV (214 nm) and liquid chromatography/electrospray ionization (LC/ESI)-MS detectors (Agilent 1260 Infinity II).The sample (150 µL) was mixed with 450 µL of denaturing buffer (6 M guanidinium hydrochloride, 5.37 mM sodium citrate in 100 mM Bis-Tris, pH 6.8) and 12 µL of 1 M 1,4-dithioerytritol, and incubated at room temperature for 1 h.The samples were then filtered through 0.2-µm filters (Whatman Mini-UniPrep syringeless filters, polytetrafluorethylene membrane) before injection on to a Jupiter C4 column (5-µm particle size, 300 Å pore size, Møller et al.: CASEIN DISSOCIATION WITH pH AND TEMPERATURE 100 × 4.6 mm, Phenomenex) kept at 40°C.The elution was conducted using 0.05% vol/vol trifluoro acetic acid (TFA) in MilliQ water (solvent A) containing 33% of solvent B (0.05% vol/vol TFA in acetonitrile) initially, and a gradient to 45.5% of solvent B after 25 min at a flow rate of 0.7 mL/min.Peaks arising from RP-HPLC analyses were identified by comparing the chromatograms with the chromatograms from LC-MS.
For MS measurements the proteins were separated on a longer Jupiter C4 column (250 × 2.00 mm, 5-µm particle size, 300 Å pores; Phenomenex) kept at 40°C, and with a different elution profile 33% of solvent B at 0 min, 44.3% after 40 min, at a flow rate of 0.3 mL/min.
For MS analysis, the Masshunter 10 Bioconfirm software (Agilent) was used to deconvolute the ion masses obtained, with the following method details: charges from 7 to 40, deconvolution mass range from 10,000 to 35,000 Da, mass step of 0.05 Da, 50% peak height for calculation of average mass, mass tolerance of ± 1.5 Da (Thesbjerg et al., 2022).The deconvoluted masses were compared with an in-house database.
Changes in Ca and P content were measured by inductively coupled plasma (ICP)-MS.Before the ICP-MS analysis, samples were acid hydrolyzed in the Multiwave 3000 (Anton Paar), using the 64MG5 rotor (Anton Paar) to ensure all P, also bound to the pro- tein, could be quantified.The ashing was carried out by mixing 30 to 50 mg of sample with 0.5 to 1 mL of concentrated HNO 3 in Wheaton glass vials (#41214).The microwave program was set to reach 800 W in 10 min, hold for 10 min, decrease to 0 W, and hold for 10 min.Temperature limit was set to 145°C.After digestion, approximately 4 mL of MilliQ water was added to achieve a final dilution factor of 150 to 200.The mass of the undiluted sample and the diluted sample after digestion was noted to calculate the exact dilution factor.The ICP-MS was carried out on an Agilent 7900 instrument, in helium mode.Calibration standard solutions were prepared at a concentration range 0.5 to 25 mg/L of pure ICP-MS grade Ca and P (Certipur).Pure yttrium solution (Certipur) was used as internal standard at 5 mg/L.
Results from all 3 analyses (RP-HPLC, LC-MS, and ICP-MS) were collected from at least 4 or 2 independent experiments for casein micelle suspensions or skim milk controls, respectively.Statistical difference (P < 0.05) between samples was determined with a Student's t-test.

RESULTS AND DISCUSSION
Figure 1 shows the elution chromatograms of skim milk (Figure 1A), full suspensions (Figure 1B) and their respective supernatants after incubation at pH 6.8 and 6.0 and at 4°C and 22°C (Figure 1C-F).Fig- ure 1A illustrates all protein identified in skim milk control.The peak assignment was carried out based on literature data (Frederiksen et al., 2011).Glycosylated (glyc) κ-CN B, E, and A with 1P (where P indicates phosphorylation) eluted the earliest, followed by κ-CN A no glyc (with 1P to 2P), and then variant B and E (also no glyc with 1-2 P).The fourth elution peak contained α S2 -CN, with 10P to 13P, followed by α S1 -CN with 8P and 9P.β-casein species B, A1, A2, and I eluted after 14 to 16 min from injection, with variant B being the small peak at 14 min.A1 eluted with the first shoulder of the large peak at 15 min and A2 and I constituted the second part of this peak.The small peak eluting right after (15.5 min) contained the γ-CN 1, 2, and 3.All β-CN species had 5P.After approximately 17 min, WP eluted in the following order: β-LG variant B, α-LA, and β-LG variant A. A comparison of Figure 1A and 1B shows that in the micellar suspension there was a smaller concentration for all casein peaks, as a result of the elimination of the soluble caseins during the first centrifugal separation (see Figure 1E) or the addition of protein-depleted permeate.As expected, the casein micelle suspension contained all the casein proteins, and only trace levels of WP (Table 1).Values were calculated as the percentual ratio of the protein area peak and that of the same peak in the corresponding suspension.Values are means (± SD) of 5 and 2 repetitions for micellar suspensions and skim milk controls, respectively.
Figure 1C and 1E show the composition of the centrifugal supernatant of skim milk incubated and separated at 4°C and 22°C, respectively and at the 2 pH values.Note the difference in scale between Figure 1A and 1B versus Figure 1C, 1D, 1E and 1F.There was a clear difference in the casein dissociation with temperature, with a higher amount of β-and α S1 -CN dissociating at 4°C and pH 6 (dotted line; Figure 1C).After incubation at 4°C and pH 6, up to approximately 33% of the total β-CN was recovered in the centrifugal supernatant, whereas significantly less, 22% of β-CN was recovered in supernatant from skim milk stored at 4°C at native pH.At 22°C there was much less dissociation, with no differences in composition with lowering the pH to 6.0 (Figure 1E).
The suspensions recombined with UF permeate contained only traces of WP (Figure 1B).Regardless, there was a large β-CN peak, corresponding to about 35% dissociated at 4°C, pH 6.0, but the other caseins' peaks (κ, α S2 , α S1 ) were comparatively lower than those in the skim milk (Figure 1C vs. Figure 1D).At 22°C, virtually no caseins were recovered in the supernatant, regardless of pH (Figure 1F).
Table 2 summarizes the changes in concentration for each casein and WP in the supernatant relative to their corresponding original suspension, calculated from peak area ratios.For suspensions and skim milk stored at 22°C, the amount of caseins remaining in the soluble phase was low, with no significant difference between the 2 pH treatments; only about 2.5% to 3% of the total casein was solubilized.A higher amount of soluble casein was recovered in the supernatants from samples incubated at 4°C, whereof the largest fraction was β-CN.Regardless of WP presence, there was a higher amount of total casein recovered at the lower pH (~6.0), namely around 11% for both suspensions and skim milk control.This agrees with previous reports (Dalgleish and Law, 1988) which reported that approximately 13% of casein was released upon skim milk storage at 4°C.In agreement with Dalgleish and Law (1988), Table 2 also demonstrates that β-CN constituted the largest fraction of released caseins at 4°C, corresponding to approximately 22% to 24% of the total β-CN, for both suspensions and skim milk control.At this cold temperature, at pH 6.8 there was little difference between samples with and without WP, and at pH 6.0, both samples contained high soluble β-CN (about 35% of the original amount).However, at these cold temperatures, when the WP were present, there was also a high recovery of caseins, with significant peaks for κ-, α S2 -and α S1 -CN.This was in contrast to the study from Dalgleish and Law (1988), who reported that the ratios of κ-, α S1 -and α S2 -CN in the serum phase did not change significantly in response to cold storage and acidification to pH 6.0.
These results suggested that at 4°C and under mildly acidic condition, WP may play a role by enhancing casein dissociation.This may suggest the formation of complexes between WP and casein under these conditions; however, more work is needed to understand such interactions, as WP are known to interact with κ-CN after heat treatment, but not during cooling at low pH.
An in-depth analysis of the distribution of the β-CN genetic variants and their phosphorylation degree in the soluble fraction of the micellar suspensions after incubation at 4°C was carried out using LC-MS (Table 2).Four genetic isoforms of β-CN predominated the suspensions: variants B, A1, A2 and I constituted each 14%, 25%, 47%, and 11% of the total β-CN pool in the suspensions.All were fully phosphorylated (5P).There were significant differences in the distribution of the isoforms in the centrifugal supernatants, regardless of pH; β-CN A2 constituted the majority of the dissociated population, approximately 60% of the total dissociated β-CN.In turn, the ratio of the variants B and A1 had decreased to ~9% and 20%, respectively.This data uniquely demonstrated that there was an uneven dissociation behavior of the various genetic β-CN variants, whereof A2 was most readily displaced from the casein micelle upon cooling.This may be due to the lower exposed hydrophobicity of A2 compared with A1, which results in a lower extent of self-assembly, as it was shown in a SAXS study (Raynes et al., 2015) which reported that β-CN A2 micelles were less prone to micellization compared with A1 micelles.The variant A2 has less surface hydrophobicity compared with A1, which is linked to the His-Pro substitution in position 67 (Raynes et al., 2015).
The micellar suspensions and their respective supernatants were also further characterized for their Ca and P content, as shown in Table 3.At native pH, regardless of temperature, approximately 30% and 50% of the Ca and P, respectively, were located in the soluble phase.These values are comparable to what is found in skim milk (Gaucheron, 2005).Upon acidification to pH 6.0, Ca and P solubilization increased, regardless of the temperature, with values of approximately 70% of the total Ca and P solubilized.This is caused by the increased proton concentration, and the decreased charge of the proteins, weakening the association of Ca and inorganic PO 4 within the micellar network (Gaucheron, 2005).These results suggested that β-CN was not affected by nanocluster dissociation.Despite the fact that up to 35% of the β-CN had moved to the soluble phase in the cold stored micellar suspensions a pH 6.0, the Ca and P distribution was unchanged with temperature, but was solely dependent of charge (pH).Furthermore, between 73% and 77% of the soluble Ca and P present in the supernatant was not associated with proteins, as previously shown (Barone et al., 2021).The internal structure of the micelles is supported by several forces that do not necessarily depend on one another, as previously discussed (Dalgleish and Law, 1989).Also, depletion of WP does not affect the nanocluster dissociation.It is clear that the stability of the casein micelle is supported by several forces that do not directly depend on one another.

CONCLUSIONS
Data showed that micellar caseins without WP underwent up to 70% colloidal calcium phosphate dissociation in response to acidification to pH 6.0.However, the increased dissociation of β-CN at low temperature (up to 35% of the total) did not lead to an increase in inorganic PO 4 and Ca release in the serum phase.
Data also suggested that WP play a role in caseins' dissociation, with an increased amount of α S -and κ-CN in the presence of WP.Moreover, it was found that A2 β-CN preferentially dissociated the micellar phase upon cooling and acidification to pH 6.0.This data showed that even small changes in treatment such as mild acidification, cooling, and presence of WP seem to have great effect on the casein dissociation, and even on the genetic composition of soluble β-CN.
Figure1.Reverse-phase HPLC elution profiles for skim milk control (A, C, E) and casein micelle suspensions with no whey protein (B, D, F).Skim milk and micellar casein suspensions (A, B), supernatants after incubation and separation at 4°C (C, D), and at 22°C (E, F), pH 6.8 (full line), pH 6.0 (dotted line).High degree of glycosylation is indicated by "glyc," and number of "P's" indicate the number of phosphorylations.Note the differences in scale between panels A and B vs. C, D, E, and F.

Table 1 .
Møller et al.: CASEIN DISSOCIATION WITH pH AND TEMPERATURE Amount of proteins remaining in the supernatant after centrifugation of the micellar phase in skim milk control or in micellar casein suspensions without whey proteins (WP) 1 Different superscript letters indicate significant difference by pairwise comparison of micellar suspension values and skim milk control values within the same row (P < 0.05, Student's t-test).

Table 2 .
Møller et al.:CASEIN DISSOCIATION WITH pH AND TEMPERATURE Distribution of fully phosphorylated (5P) β-CN genetic variants in dissociated casein micelle suspensions incubated at 4°C, at pH 6.8 and 6.0 1

Table 3 .
Calcium and phosphorus present in micellar suspensions and the corresponding soluble fractions 1 Mean values within the same column are significantly different if different superscript letter (P < 0.05, Student's t-test).1Values are means of 4 replicates, reported in mg/L ± SD.