Calcium bioaccessibility increased during gastrointestinal digestion of -lactalbumin and -lactoglobulin

Calcium bioaccessibility depends on the amount of soluble calcium under intestinal digestion. The changes in calcium during in vitro static digestion of α -lactalbumin and β -lactoglobulin in presence of calcium chloride (0 mM, 20 mM and 50 mM) were followed by combining electrochemical determination of free calcium with the determination of soluble calcium by inductively coupled plasma optical emission spectroscopy. α -Lactalbumin and, more evident, β -lactoglobulin were found to increase calcium bioaccessibility with increasing intestinal digestion time by around 5% and 10%, respectively, due to the complex binding of calcium to peptides formed from protein hydrolysis by gastrointestinal enzymes. In vitro digested samples of β -lactoglobulin in presence of CaCl 2 had nearly twice as much complex bound calcium as α -lactalbumin samples. The calcium bioaccessibility decreased significantly with the increasing concentration of added calcium chloride, although the amount of calcium chloride had little effect on the extension of digestion of α -lactalbumin and β -lactoglobulin. Simulated digestion fluids were found to have a negative effect on calcium bioaccessibility, especially the presence of hydrogen phosphate, and the amount of precipitated calcium increased significantly with increasing amount of added calcium chloride. Based on analysis and visualization by sequences of the peptides formed during digestion of α -lactalbumin and β -lactoglobulin, it was observed that peptides containing aspartic acid and glutamic acid acting as calcium chelators, may prevent precipitation of calcium in the intestines and increase calcium bioaccessibility. These results provide knowledge for the design of new dairy based functional foods to prevent calcium deficiency.


Introduction
Calcium is an essential nutrient required for biological functions in the human body such as muscle contraction and enzyme activity regulation, and a lifelong supply is important for maintaining bone strength.About 70% of the calcium in the adult diet comes from milk and dairy products in many societies (Gueguen & Pointillart, 2000;Miller et al., 2001).Calcium must be soluble, or become soluble, during digestion and remain soluble in the intestines in order to be absorbed by the human body.Calcium is primarily absorbed by transcellular transport at low calcium intakes via calcium channels and membrane bound transport proteins, and by paracellular pathway at higher calcium intakes crossing the intestinal mucosa (Bronner, 2009;de Barboza et al., 2015).Only one-third of calcium in milk is soluble, so it is important to increase calcium absorption efficiency from foods especially for people at risk of developing osteoporosis (Gaucheron, 2005).
Calcium bioaccessibility refers to the calcium potentially available for absorption in the human body and depends on soluble calcium released from the food matrix during digestion, while calcium bioavailability is the overall nutritional efficiency (Etcheverry et al., 2012).There is an increasing interest to study the availability of calcium from foods, and in vitro static methods simulating gastrointestinal behavior of nutrient absorption have provided new insights regarding the calcium speciation during the digestion of protein rich products (Wang et al., 2020).Calcium binding to ligands such as peptides or citrates will compete with precipitation in the neutral environment of the intestines, but it may lower the free calcium concentration below the concentration required for the free diffusion of calcium ions by paracellular calcium absorption (Skibsted, 2016).In our previous study, the addition of whey protein isolates to freeze-dried yogurt based snacks was found to improve calcium bioaccessibility despite a lower total calcium concentration (Wang et al., 2020).Based on these results, the hypothesis behind the present study is that calcium bioaccessibility increases along the digestion time, due to calcium binding to peptides formed from the digestion of whey proteins that prevents calcium precipitation in the intestines.Whey proteins are widely used in the food industry as ingredients because of a variety of functional properties, and α-lactalbumin (α-La) and β-lactoglobulin (β-Lg) both have good functional properties and high nutritive value.β-Lg constitutes about 50-60% of the total whey proteins, while α-La accounts for about 20% of the whey protein mass (Anandharamakrishnan et al., 2008;Foegeding et al., 2002).β-Lg is a highly pepsin-resistant protein that in its native form (i.e.no heat treatment) is unaffected during gastric digestion, whereas degradation of β-Lg occurs in the duodenal phase (Dupont et al., 2010;Kopf-Bolanz, Schwander, Gijs, Vergères, Portmann, & Egger, 2012).Milk processing had an impact on the digestibility of milk proteins, with heat treatment and fermentation showing greater sensitivity to hydrolysis of β-Lg (Kopf-Bolanz, Schwander, Gijs, Vergères, Portmann, & Egger, 2014).It has been found that calcium-binding peptides could be produced from bovine whey proteins by trypsin hydrolysis (Kim et al., 2004).A further study with Caco-2 cells has shown that whey protein hydrolysates chelated calcium resulting in higher calcium bioavailability than those found for calcium chloride and calcium gluconate under both acidic and neutral conditions, and the main calcium binding sites were identified as carboxyl and carbonyl groups of peptides (Cai et al., 2015).Calcium binding to dipeptides from hydrolysis of α-lactalbumin was found to increase in affinity with increasing pH from 5 to 9 (Jiang et al., 2021).
Further studies have shown that peptides from hydrolyzed whey have the capacity to increase calcium solubility both under acidic and neutral conditions increasing bioavailability in the gastrointestinal tract (Zhao et al., 2014).There are many other studies on the calcium binding to peptides formed by enzymatic hydrolysis, however, to the best of our knowledge, the impact of calcium binding peptides formed during the gastrointestinal digestion on calcium bioaccessibility has not been studied, despite that it simulates the process of protein ingestion in humans.
There is a limited understanding regarding calcium binding to the large amount and diversity of peptides formed from α-La or β-Lg during digestion by gastrointestinal enzymes, which hampers the design of functional foods for preventing calcium deficiency.Accordingly, the present work aims to assess the changes in calcium speciation caused by the peptides that are formed during in vitro static digestion of α-La and β-Lg, with a focus on intestinal conditions, and their impact on calcium bioaccessibility.The effect of simulated digestion fluids, which are rich in calcium binding salts, and changes in pH during the intestinal phase were also investigated.Lastly, the amino acid sequences of peptides formed from protein digestion by gastrointestinal enzymes were correlated with the changes in calcium speciation between free calcium and complex bound calcium.

Materials and chemicals
Freeze-dried α-La and β-Lg were isolated and purified from bovine milk by ultrafiltration and ion-exchange chromatography (Kristiansen et al., 1998), and stored at the University of Copenhagen at − 20 ℃.The protein content of α-La and β-Lg was more than 96% in dry matter.α-La samples contained about 9% of β-Lg, and β-Lg samples were pure without contamination with other proteins measured by SDS-PAGE as shown in Fig. S1.The calcium content of α-La and β-Lg was 2.585 ± 0.008 mg/g and 0.040 ± 0.001 mg/g, respectively.

Sample preparation
The samples of α-La and β-Lg with different concentrations of CaCl were prepared using a stock solution of 100 mM of CaCl 2 .At first, 0.60 g of α-La were dissolved in 30.0 mL Milli-Q water and then divided into six portions of 5.0 mL each.Two portions of α-La were added 5.0 mL of Milli-Q water, two portions were added 2.0 mL of CaCl 2 stock solution and 3.0 mL Milli-Q water, and the last two portions were added 5.0 mL of CaCl 2 stock solution to get the 10.0 mL samples in duplicate consisting of 0.10 g α-La (α-La), 0.10 g α-La and 20 mM CaCl 2 (α-La/Ca 20), and 0.10 g α-La and 50 mM CaCl 2 (α-La/Ca 50), respectively.The same preparation method was used for β-Lg to get the samples of β-Lg, β-Lg/Ca 20 and β-Lg/Ca 50.The concentrations of whey proteins (1.0%) and calcium (20 mM) were similar to the composition of milk (Barone et al., 2021).

In vitro static digestion
A three-step in vitro digestion was performed according to the COST INFOGEST method by Minekus et al. (2014) with minor adjustments.The simulated digestion fluids including simulated salivary fluid (SSF), simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) were daily made up as shown in Table 1 for storage at − 20 ℃, and preheated to 37 ℃ before use (Minekus et al., 2014).
To simulate the oral phase, 10.0 mL of samples were mixed with 10.0 mL of SSF and incubated for 2 min using magnetic stirring at rpm in a 37 ℃ water bath (Julabo EH-19, Seelbach, Germany) in duplicate experiments.Then, 16.0 mL of SGF were added and pH was reduced to 3.0 using HCl to simulate gastric conditions.Porcine pepsin (2000 U/mL in the final digestion mixture) was thoroughly dissolved in SGF, and 4.0 mL of pepsin solution were added to the mixture (40.0 mL in total) to initiate the experiment at time zero.No extra calcium (0.3 mol/L) was added in this step as calcium was added to the protein samples.After two hours, 15.0 mL of the gastric mixture were mixed with 10.0 mL of SIF and the pH was adjusted to 7.0 to simulate intestinal conditions.Pancreatin (trypsin activity 100 U/mL in the final digestion mixture) was dissolved in SIF, and 5.0 mL of pancreatin solution were added to the final mixture to a total of 30.0 mL.The simulated gastric and intestinal phases were both incubated for 2 h with a stirring speed of 200 rpm at 37 ℃, and pH was readjusted every 30 min during digestion.
The free calcium concentration (see below) was measured using a calcium electrode every 30 min during intestinal digestion.10.0 mL of intestinal mixtures were sampled every 1 h including at time zero and trypsin activity was inhibited using 500 µL of 1.0 mM Pefabloc.All samples were stored at − 80 ℃ for further chemical analysis.

Interactions of CaCl 2 with simulated digestion fluids
In order to investigate the impact of simulated digestion fluids on calcium bioaccessibility, the samples of 20 mM CaCl 2 (Ca 20) and mM CaCl 2 (Ca 50) without proteins were prepared in duplicate.The digestion with simulated digestion fluids followed the same procedure as described in Section 2.3, and for the digestion of CaCl 2 without

Table 1
Preparation of simulated digestion fluids (SSF-simulated salivary fluid, SGFsimulated gastric fluid and SIF-simulated intestinal fluid) according to the COST INFOGEST protocol (Minekus et al., 2014) simulated digestion fluids (nf), Milli-Q water was added instead of SSF, SGF and SIF.The concentration of pepsin and pancreatin were the same as described in Section 2.3.During the intestinal phase, the free calcium (see below) was quantified every 30 min and 10.0 mL aliquots were taken every hour for further analysis.

Electrochemical analysis of free calcium (Ca 2+ )
Calcium ion activity was measured using a multimeter (sensION + MM374, Hach lange, Barcelona, Spain) combined with a calcium selective electrode ISE25Ca and a reference electrode REF251 (Radiometer, Copenhagen, Denmark).Electrodes were calibrated at 37 ℃ before measurement using calcium chloride standard solutions with concentrations of 0.01000, 0.1000, 1.000 and 10.00 mM.A linear standard curve was obtained according to the relation between the electric potential (mV) and − log (a Ca2+ ) in agreement with Nernst equation (de Zawadzki & Skibsted, 2020).This calibration based on thermodynamics makes the calibration valid in any aqueous medium since all ions present in each solution are included in calculation of the ionic strength.Calcium ion activity in the standard solutions was calculated according to: where a Ca2+ is the ion activity of calcium and γ Ca2+ is the coefficient of activity based on Davies' equation (Davies, 1962): where A DH is 0.519 (Debye-Hückel constant at 37 ℃) and z is the calcium ion charge of 2. I is the ionic strength calculated according to the concentration of all ions of the electrolytes present in the solutions: where c i is the concentration of ions in the simulated digestion fluids, and z i is the corresponding charge.

Inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis
Calcium content of α-La and β-Lg was analyzed in duplicate using inductively coupled plasma optical emission spectroscopy (5100 ICP-OES, Agilent Technologies, Santa Clara, USA) at the multiple wavelengths of 393.366, 396.847 and 422.673 nm (de Zawadzki & Skibsted, 2020).A standard curve was obtained from calcium chloride solutions with the concentration from 0.00 to 10.0 mg/L.For the powder samples, microwave digestion (Rotor 12HVT50 in Multiwave GO, Anton Paar, Graz, Austria) prior to analysis was necessary to ensure that all the organic matter was dissolved.Briefly, 0.200 g of α-La or β-Lg were diluted by 8.0 mL of 65% HNO 3 (Emplura®), 2.0 mL of 37% HCl (Emprove®) and 2.0 mL of H 2 O 2 (95321) in HVT50 vessels followed by a digestion program at 100 ℃ for 2 min and then 180 ℃ for 8 min.The chemicals used for microwave digestion were obtained from Sigma-Aldrich (St. Louis, USA).Calcium content was analyzed by ICP-OES after dilution by Milli-Q water to 3-5% acid.
Soluble calcium content was determined using the samples taken from in vitro intestinal digestion at 0 h, 1 h and 2 h in Section 2.3.All the samples were ultrafiltrated at 5000 rpm for 20 min at 4 ℃ with centrifugal ultrafilters of molecular weight cut off 3 kDa (VIVASPIN 20, Sartorius Stedim Lab Ltd, Stonehouse, UK) to get the ultrafiltrate (Lorieau et al., 2018).Ultrafiltrated samples were diluted 10 times with 5% of HNO 3 (Suprapur® Nitric acid 65%, Merck KGaA, Darmstadt, Germany) and soluble calcium concentration was subsequently measured by ICP-OES analysis (Wang et al., 2020).Precipitated calcium remaining after soluble calcium was removed from the total calcium content.
Complex bound calcium is the difference between soluble calcium and free calcium, which was calculated according to (Walser, 1961): Calcium bioaccessibility was calculated according to: calcium bioaccessibility (%) = soluble calcium (mM) total calcium in samples (mM) × 100 (5)

Hydrolysis of β-lactoglobulin by pancreatin
The binding between calcium and peptides of β-Lg formed from pancreatin hydrolysis was investigated by a simplified model of intestinal digestion in duplicate at 37 ℃.Samples of 25.0 mL with 0.050 g of β-Lg and 1.0 mM of CaCl 2 were prepared followed by adding 5.0 mL of pancreatin solution (trypsin activity 100 U/mL in the final digestion mixture).The pH of samples and enzyme solution was adjusted to 7.0, 8.0 or 9.0 using dropwise addition of NaOH solution in agreement with the pH range for pancreatin.These pH values were selected as the pH in the intestinal phase increase to 8.5 because of the presence of (NH4) 2 CO 3 in the simulated intestinal fluids.All the solutions were prepared using Milli-Q water.Free calcium concentration, electrical conductivity and pH were detected as the reaction progressed.Electrical conductivity was determined using a multimeter (sensION + EC71, Hach lange, Barcelona, Spain) with a conductivity cell (Sension + 5070, Mettler Toledo, Schwerzenbach, Switzerland) (Wang et al., 2020).

Peptide analysis by LC-MS/MS
Samples of 0.5 mL collected from in vitro intestinal digestion (Section 2.3) were acidified with 5.0 µL of formic acid (98-100%, Fisher Scientific UK, Loughborough, UK) followed by filtration of 0.22 µm filters (Q-Max syringe filters, Frisenette ApS, Knebel, Denmark).The peptides were analyzed in duplicate.Briefly, samples were separated in a bio-Zen™ 1.7 µm Peptide XB-C18 column (150 × 2.1 mm, Phenomenex, Cheshire, UK) using an UltiMate 3000 UHPLC system coupled with a Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, USA) as described by Aalaei et al. (2021).Samples of 10 µL were injected at a flow rate of 0.25 mL/min and eluted with 100% solvent A (0.10% formic acid in Milli-Q water) for 5 min, followed by a gradient of 0 to 50% solvent B (0.10% formic acid in 80% acetonitrile) for 40 min.The mass spectrometry was carried out using data dependent MS/MS.The LC-MS/ MS data were processed and compared with the whole bovine genome from Swiss-Prot database using Proteome Discoverer 2.2 (Thermo Fisher Scientific, Roskilde, Denmark).For the visualization of precursor proteins, Peptigram (http://bioware.ucd.ie/peptigram/) was used to compare multiple samples of the peptide sequences and intensities (Manguy et al., 2017).

Statistical analysis
There were two experiments carried out independently.The data obtained were analyzed using one-way ANOVA by SPSS Statistics 27 (SPSS Inc., Chicago, USA).The results were expressed as mean ± standard deviation, and a value of P < 0.05 indicated a statistically significant difference by Tukey's test.

Calcium speciation during in vitro digestion of α-La and β-Lg in the presence of CaCl 2
Table 2 shows the calcium concentration changes during the in vitro intestinal digestion of α-La and β-Lg.The total calcium concentration in each sample was calculated based on the calcium content of proteins and the amount of calcium chloride added, which was constant during in vitro digestion.β-Lg contained only a small amount of calcium and consequently the calcium present in the samples was mainly due to the added calcium.The use of calcium concentration of 20 mM was similar to the composition of milk, and 50 mM of calcium chloride was added to investigate the potential for extra calcium and peptide binding, at concentrations relevant for high calcium fortification of milk.For samples with the addition of 20 mM and 50 mM of calcium chloride, the concentration of calcium present in the intestinal phase of digestion due to dilution was 2.50 mM and 6.25 mM, respectively.
Soluble calcium was determined by ICP-OES in the ultrafiltrates sampled at 0 h, 1 h and 2 h during the in vitro intestinal digestion.Any precipitation with a molar mass higher than 3 kDa is expected to be unable to pass the ultrafilters, which should not be a problem as the highest calcium binding capacity was expected for peptide fragments with a molar mass<2 kDa (Liu et al., 2013).For the samples with α-La, soluble calcium increased insignificantly by 4.9% (α-La/Ca 20) and 5.5% (α-La/Ca 50) as intestinal digestion proceeded, probably due to the competition between calcium binding to peptides formed during digestion and precipitation of calcium by the simulated digestion fluids.For the samples with β-Lg, soluble calcium increased significantly by 10% (β-Lg/Ca 20) and 13% (β-Lg/Ca 50) as the intestinal digestion proceeded.Moreover, the soluble calcium from β-Lg/Ca 20 was increased by 9% from 0 h to 1 h and only by 1% from 1 h to 2 h of the intestinal digestion.The concentration of soluble calcium during the intestinal digestion of β-Lg/Ca 50 increased by 11% and 2% from 0 h to 1 h and from 1 h to 2 h, respectively, indicating that β-Lg was hydrolyzed to produce peptides primarily at the beginning of the intestinal digestion.Soluble calcium increased while precipitated calcium decreased during intestinal digestion.The amount of precipitated calcium increased significantly by a factor of 4.3 for both α-La and β-Lg with higher calcium chloride addition (50 mM) at the end of digestion.Soluble calcium is the sum of free calcium and complex bound calcium (Walser, 1961), and therefore the changes in both free calcium and complex bound calcium were analyzed for a better understanding of the changes in speciation.
Free calcium as ionic calcium was determined electrochemically during the intestinal phase of digestion.As intestinal digestion proceeded, the free calcium concentration of α-La/Ca 20 and α-La/Ca 50 solutions was reduced significantly by 8% and 19%, respectively.At pH 7, free calcium supplied by calcium chloride decreased with digestion time because peptides formed by hydrolysis of the α-La were bound to calcium and consequently reduced the amount of free calcium.The fluctuations in the free calcium content of the samples with only α-La were explained by very small measured values close to the detection limit of the calcium electrode.In β-Lg samples, free calcium decreased even more by 18% for β-Lg/Ca 20 and 32% for β-Lg/Ca 50 from the beginning to the end of the intestinal digestion.It has been demonstrated that calcium binding by ligands, such as proteins, peptides, and hydroxycarboxylates competes with precipitation at neutral pH and may enhance solubility, although the formation of strong complexes with calcium also will reduce the free calcium concentration (Perego et al., 2015;Skibsted, 2016).Precipitation of calcium with components of the simulated digestion fluids seems responsible for the difference between the free calcium decrease and the soluble calcium increase.The samples of β-Lg without calcium addition had calcium concentration below the detection limit of the calcium electrode, as no free calcium could be detected.
Complex bound calcium increased from the beginning to the end of intestinal digestion.Soluble calcium concentration was higher than free calcium concentration in all samples because calcium was bound to peptides and salt anions as complexes and not detected by the electrode although soluble (Lorieau et al., 2018).Samples with the higher additional calcium content of 50 mM had a concentration of complex bound calcium higher by a factor of 2.7 compared to the samples with 20 mM of calcium addition at the end of the digestion (Table 2), a difference that is explained by calcium binding to peptides formed from α-La and β-Lg having larger numbers of calcium binding sites.Calcium ions can form complexes via amino groups and carboxylic groups of peptides (Liu et al., 2013).Compared to α-La/Ca 20 and α-La/Ca 50, calcium complexes in β-Lg/Ca 20 and β-Lg/Ca 50 were almost present at a double concentration at all sampling times.Samples with β-Lg may be digested to produce a higher concentration of peptides binding calcium, in effect resulting in a lower free calcium concentration and a higher complex bound calcium content as compared to α-La.The ratio between complex bound calcium and soluble calcium (complex/soluble) increased from 5% to 17% (α-La/Ca 20), from 1% to 24% (α-La/Ca 50), from 9% to 32% (β-Lg/Ca 20), and from 2% to 41% (β-Lg/Ca 50), respectively, as intestinal digestion continued.The initial complex/soluble ratio for the samples with 50 mM of calcium addition was lower than the ratio for the other samples and could be due to a slow release of peptides in the presence of the massive free calcium concentration at the beginning of intestinal digestion.
Calcium bioaccessibility is the calcium in the soluble form which is potentially available for absorption in the intestines (Etcheverry et al., 2012), and has been found to be dependent on the concentration of total calcium as shown in Fig. 1.Samples with 50 mM of calcium chloride had lower calcium bioaccessibility compared to samples with 20 mM of calcium chloride, despite higher soluble calcium and total calcium Different superscript letters in the same column represent significant differences with intestinal digestion time in Tukey (p < 0.05)."\" = not detected.
content, because the increase in soluble calcium was lower than the increase in total calcium resulting from the addition of calcium chloride.
In the present study, calcium bioaccessibility increased with digestion time by 5.4% for α-La/Ca 20, 5.2% for α-La/Ca 50, 9.3% for β-Lg/Ca 20 and 13.1% for β-Lg/Ca 50.β-Lg enhanced calcium bioaccessibility more than α-La in the samples with calcium addition due to the increase in complex bound calcium by peptides formed by gastrointestinal digestion of β-Lg.Calcium absorption by the human body depends on the content of soluble calcium in the intestines, and calcium bound to peptides can prevent calcium precipitation by complexation at neutral pH and consequently increase calcium bioaccessibility (Perego et al., 2015;Sun et al., 2017).Calcium binding to peptides and small proteins is coordinative and non-covalent, and calcium equilibria in aqueous systems are adjusting fast under neutral and acidic conditions with a few exceptions (Garcia et al., 2020).Accordingly, calcium bound to the peptides will dissociate when calcium is absorbed in the intestines through one of the two absorption mechanisms: paracellular or transcellular pathway.

Interactions of CaCl 2 with simulated digestion fluids
The simulated digestion fluids used in the intestinal digestion protocol contained 5.0 mL of simulated salivary fluid (SSF), 10.0 mL of simulated gastric fluid (SGF) and 15.0 mL of simulated intestinal fluid (SIF).All the simulated digestion fluids contained phosphate and  carbonate (Minekus et al., 2014), which could bind calcium either in complexes or in precipitates.In order to study the influence of these electrolytes on calcium bioaccessibility, a control experiment consisting of calcium chloride solutions with or without simulated digestion fluids was subjected to the in vitro digestion.
The soluble calcium concentration of Ca 20 and Ca 50 solutions increased slightly but not significantly as the intestinal digestion proceeded, while free calcium decreased and complex bound calcium increased significantly as seen in Fig. 2. For the samples digested with Milli-Q water (nf -no simulated digestion fluids), soluble calcium of Ca 20 nf and Ca 50 nf was almost constant and close to the total calcium content.Therefore, calcium bioaccessibility of Ca 20 nf and Ca 50 nf ranged from 95% to 97%, in agreement with the observation that the ultrafiltrates recovered almost all the calcium in the soluble form.There was a slight increase in free calcium for Ca 20 nf and for Ca 50 nf corresponding to a small, hardly significant decrease in complex bound calcium with time.
Simulated digestion fluids showed little effect on the soluble calcium concentration at a lower calcium chloride (20 mM) addition, however, digestion fluids decreased soluble calcium of Ca 50 by 26% compared to Ca 50 nf at the end of digestion, which was consistent with the observed calcium bioaccessibility.The simulated digestion fluids caused precipitation at a higher concentration of free calcium.At the high calcium concentration, the ionic product of the calcium salt and some components of the simulated digestion fluids becomes larger than the solubility product, causing supersaturation of the solution leading to precipitation.This will not happen at the lower calcium concentration.It has been found that the presence of simulated digestive fluids, enzymes and bile salts had a negative effect on zinc bioaccessibility during in vitro static digestion (Rousseau et al., 2019).Free calcium concentration decreased with the presence of simulated digestion fluids by a factor of 1.5 at 2 h of intestinal digestion due to the combination of calcium and other electrolytes with the potential of forming precipitates.Complex binding is instantaneous while precipitation is slower (Carr & Swartzfager, 1975).At pH 7.0, phosphates were present as 50% of H 2 PO 4 -and 50% of HPO 4 2-, and carbonate is reported to be distributed between 20% of H 2 CO 3 and 80% of HCO 3 -in simulated digestion fluids (Pismenskaya et al., 2001).
Hydrogen phosphate with a binding constant of 681 L/mol at 37 ℃ binds calcium far stronger than dihydrogen phosphate and bicarbonate, and CaHCO 3 + complex is reported to be found to be fully ultrafiltrable at pH 7.3 (Chughtai et al., 1968;Pedersen, 1971).Therefore, the decrease of free calcium and the increase of complex bound calcium can be assigned quantitatively to complexation with phosphates, mainly as hydrogen phosphate, and with bicarbonate.

Effect of intestinal pH on hydrolysis of β-Lg by pancreatin
From the calcium results in Fig. 1, it is observed that β-Lg improved calcium bioaccessibility more than α-La, which may be related to a higher degree of complexation of calcium by peptides formed during intestinal digestion.Furthermore, β-Lg is resistant to pepsin digestion under gastric conditions, and free amino acids are mainly generated in the intestinal phase (Egger et al., 2016;Hodgkinson et al., 2018).There is no calcium associated with β-Lg below the protein isoelectric pH of 5.2 (Patocka & Jelen, 1991).Therefore, a simplified β-Lg hydrolysis trial with pancreatin was performed to clarify the effect of intestinal pHs that increases in the intestinal phase to 8.5 due to the presence of (NH4) 2 CO 3 in the simulated intestinal fluids.Fig. 3a shows the changes in free calcium during hydrolysis of β-Lg by pancreatin in the presence of 1.0 mM of calcium chloride at 37 ℃.Free calcium concentration increased during the first 0.5 h, then decreased for up to 1.5 h and then stayed stable with minor fluctuations.The initial increase in free calcium concentration seems related to the initial decrease in pH during early hydrolysis leading to weaker binding of calcium due to increasing protein protonation.From the first time point (2 min) to 1.5 h, free calcium concentration was reduced by 13% at pH 7, by 19% at pH 8 and by 20% at pH 9, indicating that peptides produced by pancreatin hydrolysis bound calcium and accordingly decreased free calcium concentration.As the intestinal pH increased, the hydrolysis of β-Lg by pancreatin was enhanced to produce more peptides binding calcium.Meanwhile, free calcium decreased even more at higher pH because calcium binding affinity increases with increasing pH (Tang & Skibsted, 2016).In Fig. 3b, the increasing conductivity by around 26% in all samples also confirmed that peptide charge increased with the formation of peptides and amino acids for the pH intervals studied.At 2 min, pH decreased rapidly to 7.96 ± 0.06 (from pH 9), to 7.33 ± 0.01 (from pH 8) and to 6.87 ± 0.02 (from pH 7) as seen in Fig. 3c.The pH continued to decrease slowly during the next 1.5 h consistent with the free calcium changes and in agreement with the continuing proteolysis.Hydrolysis of proteins decreased pH and subsequently liberated calcium ions, however, the complex binding between hydrolysates of β-Lg and calcium ions reduced the free calcium at the same time.This observation confirmed that the decrease in free calcium content was caused by the complex binding of calcium to peptides rather than precipitation due to the similar final pH of samples ranging between 6.25 and 6.71.The results indicated that increased pH from 7 to 9 under the intestinal conditions promoted the hydrolysis of β-Lg, which further protected calcium from being precipitated.

Peptides formed during in vitro digestion of α-La and β-Lg combined with CaCl 2
Further studies were performed using LC-MS/MS to identify the peptides formed from α-La and β-Lg in the presence of CaCl 2 at 0 h, 1 h and 2 h during the intestinal digestion.Only the data with high or at least medium confidence of identifications were included in the analysis.59 peptides of α-La (P00711) and 491 peptides of β-Lg (P02754) were identified.
Peptide length distribution can be used to assess the extent of protein digestibility, as more efficient digestion will produce more short peptides, thus peptides were grouped based on their length of amino acids (AAs), in 6-10 AAs, 11-15 AAs, 16-25 AAs, 26-35 AAs and 36-45 AAs (Aalaei et al., 2021;Zenker et al., 2020).The relative count was expressed as the percentage of peptide count in each length range relative to the total count of peptides in each sample as shown in Fig. 4. Peptides with 6-10 AAs were the largest count with more than 65% in all samples of α-La, and increased as intestinal digestion progressed in Fig. 4a.There was no peptide in groups from 26 to 35 AAs and 36-45 AAs identified, and the longest peptide detected in α-La was 17 AAs.From 0 h to 1 h of intestinal digestion of α-La, there was a decrease by around 16% in the size range of 11-15 AAs, and it remained the same for a longer digestion time.For the peptides of α-La with 16-25 AAs, the relative count mainly decreased by about 41% in 2 h, which indicates that longer peptides take a longer time for digestion.The relative count of β-Lg was higher in the size range of 11-15 AAs as seen in Fig. 4b.
Overall, α-La was more easily digested than β-Lg in the intestines after gastric pepsin hydrolysis.The different calcium chloride concentrations had little effect on the peptide size distribution of α-La and β-Lg during in vitro intestinal digestion.
The total count of peptides in each sample declined with intestinal digestion due to the polypeptides being further digested to short oligopeptides or free amino acids, which were not detected by LC-MS/MS.At the same time, the concentration of complex bound calcium increased as intestinal digestion proceeded.Therefore, a negative correlation was observed between complex bound calcium and the detected peptides as shown in Fig. 5.The absolute value of the correlation coefficient increased with the increase of calcium chloride presence suggesting that peptides formed potential sites for calcium binding in excess.Samples with α-La had a larger absolute value of correlation coefficient than samples with β-Lg, which moreover indicates that peptides formed from α-La may have relatively high calcium binding affinity despite a smaller count.Cyclic peptides have been designed to model the calcium binding loop of α-La, and it was found that the sequence of KFLDDLTDD was well conserved among different peptides and participated in calcium binding (Farkas et al., 2005).The main binding sites were carboxylates of aspartic acid and carbonyl oxygen atoms of lysine and aspartic acid (Farkas et al., 2005).The amino acid sequence 98-107 (KFLDDDLTDD) was also identified in our samples of α-La, and considered to contribute primarily to complex calcium binding.The abundance of KFLDDDLTDD decreased by around 55% with digestion time under intestinal conditions.Two amino acids of aspartic acid and glutamic acid had the strongest calcium binding through the carboxylate groups as calculated by density functional theory (DFT) (Tang & Skibsted, 2016).Aspartic acid and glutamic acid may interact in the fragment of 145-147 (DDE) of β-Lg, and it was speculated that this fragment had a higher calcium binding affinity considering the primary structure of peptides formed.The amino acid sequence 141-151 (TPEVDDEALEK) was the most abundant of β-Lg among the peptides sequenced including DDE fragment, whose intensity increased as intestinal digestion proceeded.
The peptide profile of α-La and β-Lg during intestinal digestion has been shown in Fig. 6 to illustrate peptide coverage, and the height and the color of bars are proportional to the count and summed intensities of peptides overlapping this region, respectively.The dashed lines indicate that there are no peptides sequenced at this position (Manguy et al., 2017).Peptide coverages of α-La and β-Lg were 68% and 91%, and the maximal intensity of α-La and β-Lg was 3.45 × 10 8 and 7.64 × 10 8 .Higher and darker bars were observed in β-Lg (Fig. 6b) than in α-La  (Fig. 6a), which showed that β-Lg had more variations of peptide count and abundance in agreement with the calcium results as shown in Table 2 and Fig. 1.It has been found that β-Lg produced more fragments per molecule than α-La after in silico digestion (Barati et al., 2020).
Smaller peptides bound relatively more calcium than large peptides, and consequently increased the soluble calcium concentration and calcium bioaccessibility in the intestines.It is important to take the limitation of analysis by LC-MS/MS into account, since only polypeptides above tetrapeptides are detected, while the short peptides and free amino acids, which also bind calcium, cannot be sequenced by LC-MS/MS.With the increasing digestion time from 0 h to 2 h, the peptide count and intensity decreased substantially in all samples analyzed probably because the polypeptides were digested into short peptides or free amino acids that could not be identified.
In Peptigram patterns of α-La (Fig. 6a), the high and dark bars indicated the high count and intensity of peptides with the fragment of 34-39 (LKGYGG).After one hour of digestion, the amino acid sequence 108-111 (IMCV) disappeared due to the further hydrolysis of α-La in this region forming free amino acids and short peptides, which could not be detected.A similar situation was found in the region with the amino acid sequence of 20-28 (EQLTKCEVF), from which the peptides were no longer detectable after two hours of digestion suggesting that they were completely hydrolyzed in two hours by pancreatin.The peptide profile of β-Lg in Fig. 6b shows that the high count and intensity of peptides were in the regions of 22-30 (TMKGLDIQK), 62-70 (LKPTPEGDL) and 141-149 (TPEVDDEAL), and the count decreased gradually with digestion time showing that very short oligopeptides were formed in the intestines.The peptide of 141-149 (TPEVDDEAL) has also been sequenced in the heavy fragment of β-Lg using in silico model (Barati et al., 2020).Short oligopeptides have synergistic effects with higher calcium binding affinity, which could increase calcium binding and subsequently enhance calcium bioaccessibility under neutral or alkaline conditions (Tang & Skibsted, 2016;Zhao et al., 2014).Further experimental studies combined with DFT calculations may elucidate calcium binding affinity of the sequenced peptides.

Conclusions
Calcium binding to peptides formed during in vitro intestinal digestion of α-La and β-Lg increased the concentration of calcium complexes to improve calcium bioaccessibility.For the same digestion time, the diversity and amount of peptides from digested β-Lg bound almost twice as much calcium as the ones from α-La at intestinal conditions.Calcium complex binding to peptides formed during digestion increased soluble calcium concentration and accordingly enhanced calcium bioaccessibility.Anions mainly hydrogen phosphate in the simulated digestion fluids, especially at higher calcium concentrations, may precipitate calcium with a negative effect on calcium bioaccessibility.An increasing pH promoted hydrolysis of β-Lg by pancreatin, which was found to protect calcium from precipitation during intestinal digestion.Peptide profiles following digestion of whey proteins showed that β-Lg formed a higher amount and diversity of peptides than α-La resulting in a higher concentration of peptides available for calcium complex formation.β-Lg accordingly increased calcium solubilization under intestinal conditions better than α-La.However, the peptides formed from α-La may have relatively high calcium binding affinity despite the smaller amount detected.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Table 2
Calcium concentration changes during in vitro intestinal digestion.