Cows' milk is widely used as a substitute for human milk, however donkey milk feeding can be an alternative in cases of food intolerance due to its resemblance to human milk (Iacono et al. Reference Iacono, Carroccio, Cavataio, Montalto, Soresi and Balsamo1992; Carroccio et al. Reference Carroccio, Cavataio, Montalto, D'Amico, Alabrese and Iacono2000; Criscione et al. Reference Criscione, Consolo, Bordonaro, Guastella, Saletti, Zuccaro, D'Urso and Marletta2009).
Cows' and donkey milks show different protein composition, with cows' milk containing 3·0–3·8 g/100 ml in contrast to donkey milk which contains 1·5–1·8 g/100 ml (Grupta, Reference Grupta1983; Andrews et al. Reference Andrews, Taylor and Owen1985; Guo et al. Reference Guo, Pang, Zhang, Zhao, Chen, Dong and Ren2007; Uniacke-Lowe et al. Reference Uniacke-Lowe, Huppertz and Fox2010; D'Alessandro et al. Reference D'Alessandro, Martemucci, Jirillo and Leo2011). Although some protein components of donkey and cows' milk are similar, sequence homology, concentrations and thermal stability are often different.
Great difference is observed in the proportion of whey proteins. In donkey milk whey proteins constitute around 50 and 60% of the nitrogen fraction, whereas in cows' milk they account for only 20% of the nitrogen fraction (Ferreira et al. Reference Ferreira, Mendes and Ferreira2001; Veloso et al. Reference Veloso, Teixeira, Peres, Mendonça and Ferreira2004; Cunsolo et al. Reference Cunsolo, Saletti, Muccilli and Foti2007). Additionally, caseins are the largest class of proteins in cows' milk (80% of total milk protein) whereas in donkey milk the caseins content is much lower. The most important allergens of cows' milk are caseins and β-lactoglobulin (β-lg) fraction (Santos & Ferreira, Reference Santos and Ferreira2007; Polidori et al. Reference Polidori, Beghelli, Mariani and Vincenzetti2009).
Several methods are used to analyse the protein fractions, namely, electrophoretic techniques (Carretero et al. Reference Carretero, Trujillo, Mor-Mur, Pla and Guamis1994; Park & Jin, Reference Park and Jin1998), isoelectric focusing (IEF) (Miralles et al. Reference Miralles, Krause, Ramos and Amigo2006; Criscione et al. Reference Criscione, Consolo, Bordonaro, Guastella, Saletti, Zuccaro, D'Urso and Marletta2009; Giambra et al. Reference Giambra, Jäger and Erhardt2010) and capillary electrophoresis (De Block et al. Reference De Block, Merchiers, Mortier, Braekman and Ooghe2003; Clément et al. Reference Clément, Agboola and Bencini2006; Criscione et al. Reference Criscione, Consolo, Bordonaro, Guastella, Saletti, Zuccaro, D'Urso and Marletta2009). High-performance liquid chromatography (HPLC) by ion-exchange (Andrews et al. Reference Andrews, Taylor and Owen1985), hydrophobic interactions (Goheen & Gibbins, Reference Goheen and Gibbins2000; Ferreira et al. Reference Ferreira, Mendes and Ferreira2001), gel filtration (Grupta, Reference Grupta1983), reversed phase (Ferreira & Caçote, Reference Ferreira and Caçote2003; Polidori et al. Reference Polidori, Beghelli, Mariani and Vincenzetti2009) and immunological methods (Haza et al. Reference Haza, Morales, Martin, Garcia, Anguita, Gonzalez and Sanz1996). Every method has its own advantages, but the use of HPLC for the analyses of native milk proteins has resulted in the development of rapid and automated analyses, characterised by good separation, high accuracy and reproducible results. The size, electrophoretic charge and polarity of the various milk proteins can be affected by heat treatment (Grupta, Reference Grupta1983). Therefore, the above mentioned methods used to separate and quantify native proteins present different strengths and weakness when used for quantification of denatured proteins which are present in thermally processed milk samples (Ferreira et al. Reference Ferreira, Mendes and Ferreira2001; Ferreira, Reference Ferreira2007).
Heat-induced denaturation and interactions of cows' milk whey proteins have been studied in different milk protein systems under a variety of experimental systems, however studies concerning thermal stability of donkey milk proteins are scarce. Methods that enable simultaneous quantification of cows' and donkey milk major proteins in native and denatured states are of major importance for routine studies of the effects of thermal processing on protein denaturation.
In the present work, cows' and donkey milks (raw and thermally processed) and respective whey were analysed for quantification of major proteins. Two different chromatographic approaches, size exclusion (SE-HPLC) and reversed-phase high performance liquid chromatography (RP-HPLC) both coupled to UV detection were used. The first chromatographic method highlights whether thermal processing causes changes of protein molecular weight such as protein aggregation and the second method shows protein denaturation that results in peak disappearance owing to changes of protein polarity. The goal of this work is to combine these two chromatographic approaches to understand protein denaturation and aggregation, in order to acknowledge the usefulness of these methods for routine control of thermal processing effect on quantitative and qualitative profile of major proteins.
Materials and Methods
Reagents and proteins standards
All reagents used were of analytical grade purity. Eluents for HPLC were filtered through 0·22 μm NL 17 filters and degassed under vacuum for at least 15 min before use. Casein from bovine milk with a minimum purity of 75% was supplied by Sigma Chemical Co (St, Louis, MO). Purified bovine standards of α-casein (α-cn), β-casein (β-cn), κ-casein (κ-cn), β-lg, α-lactalbumin (α-la), bovine serum albumin (BSA), and egg lysozyme (lys) were also supplied by Sigma Chemical Co. The molecular weight markers used were: Phosphorylase b, 97·0 kDa; Albumin, 66·0 kDa; Ovalbumin, 45·0 kDa; Carbonic anhydrase, 30·0 kDa; Trypsin inhibitor, 20·1 kDa; α-la, 14·4 kDa (LMW Calibration Kit for SDS Electrophoresis, GE Healthcare, Amersham, UK).
Sampling and sample preparation
Bulk milk obtained from 10 Holstein Friesian cows and from 10 Ragusana breed asses was used, these animals were machine milked. Milk samples were frozen immediately after collection and stored at −20°C until use. They were thawed at room temperature and shaken before their analysis.
Laboratorial thermally processed milk samples were prepared by heating raw cows' and donkey milk samples at 100°C in a water bath for 5 min and cooled in ice down to room temperature. This procedure was performed in triplicate for each milk type.
Two commercial thermally processed cows' milk samples (from different brands) that suffered ultra-high temperature processing (UHT) were also analysed.
The total protein concentration of whole milk (raw and thermally processed) and whey were determined following the method of Bradford (Bradford, Reference Bradford1976). Protein identification was performed by SDS-polyacrylamide gel electrophoresis.
For HPLC analyses milk samples were skimmed by centrifugation at 9000 g, at 4°C, for 15 min, the lipid layer was removed by siphoning with a pipette. Analyses of raw and thermally processed milks were performed after dilution with water. Whey proteins were obtained from skimmed milk by adjusting the pH to 4·6 with 10% (v/v) acetic acid and centrifuged at 9000 g, 4°C for 10 min in order to obtain a supernatant of whey proteins and remove the isoelectrically precipitated caseins.
Sodium dodecyl sulphate polyacrylamide gel electrophoresis analysis
SDS–PAGE was done using a 15% acrylamide-bis acrylamide solution and the Multiple Gel Casters Hoefer® (Holliston, MA), Rect. Glass Plates 10×10 cm Hoefer®, Mightly Small II For 8×9 cm Gels Hoefer® apparatus. The proteins were visualized on the gel by Coommassie blue 0.125%.
Chromatographic approaches
Two different chromatographic approaches were used for quantification of major proteins in cows' and donkey milks using SE-HPLC (first approach) and RP-HPLC (second approach).
The SE-HPLC equipment consisted of a Gilson chromatograph (Gilson Medical Electronics, France) equipped with a type 302 pump, a type 305 pump and a type 7125 Rheodyne Injector with a 100 μl loop. A Gilson 118 variable-wavelength ultra violet detector was used. The equipment was controlled by Gilson 712 software that controlled the solvent gradient, data acquisition and data processing. Chromatographic separation was performed on a Superdex 7510/300 GL column from Tricorn (Amersham Biosciences, UK), appropriate for separation of molecules presenting molecular weight between 3 and 70 kDa. The mobile phase was an aqueous solution containing 0·075 m NaCl and 0·025 m KH2PO4 adjusted to pH 7·0 with 1 m KOH. The flow rate was 0·5 ml/min. The effluent was monitored at 280 nm.
The reversed-phase chromatographic analysis was carried out in an analytical HPLC unit (Jasco, Tokyo, Japan) equipped with Jasco PU-2089 Plus quaternary low pressure gradient HPLC pump, a MD-2010 Plus multiwavelength detector and a 7125 Rheodyne injector valve (California, USA) with a 20 μl loop. The column was a Chrompack P 300 RP column (Sao Paulo, Brazil) that contains polystyrene-divinylbenzene copolymer-based packing (8 μm, 300 Å, 150×4·6 I.D.). A Chrompack P RP (24×4·6 mm I.D.) was used as a pre-column. The temperature of the column was adjusted and maintained at 45°C and the flow rate was 0·5 ml/min. Eluted peaks were detected at 214 nm. A Borwin PDA Controller Software (JMBS Developments, Le Fontanil, France) was also used.
Gradient elution was carried out with a mixture of two solvents. Solvent A consisted of 0·1% trifluoroacetic acid (TFA) in water and solvent B was 0·1% TFA in acetonitrile. Elution was achieved by the following gradient: 0–5 min, 33–35% B, 5–9 min, 35–37% B, 9–18 min, 37–39% B, 18–33 min, 39% B, 33–45 min, 39–55% B, 45–50 min, 55–33% B, 50–55 min, 33% B.
Quantification of major cows' and donkey milk proteins
For the quantitative determination of β-lg, α-la, lys and total casein (cn), milk samples were analysed by SE-HPLC and by RP-HPLC. Each standard solution of bovine milk β-lg, bovine milk α-la, and egg white lys, was prepared in the following concentrations: 0·25, 0·50, 1·0, 1·5, 2·0 mg/ml. Standard solutions from bovine casein were prepared in the concentrations of 1·0, 4·0, 5·5, 8·0 mg/ml (concentrations of standard solutions were always corrected according to the standard purity).
The validation of SE-HPLC and RP-HPLC methods for the quantification of major milk proteins was accomplished by testing the linearity, the detection limit, the precision (repeatability and reproducibility) and the accuracy. The linearity of the method was checked through the calibration curves, which were calculated for major proteins, and obtained by linear regression of the peak area versus concentration of each protein in the injected solution. The detection limit values (LOD) were calculated as the concentration corresponding to three times the sd of the background noise.
The precision of SE-HPLC and RP-HPLC methods was evaluated by estimating the repeatability and reproducibility. Areas under peaks of the chromatograms and retention times were used for validation purposes. One sample of raw cows' milk was analysed daily, repeating the analysis over 3 days. The reproducibility was calculated as the Relative Standard Deviation (RSD) of peaks area and retention times across days. The repeatability was studied by running 5 consecutive replications of the same sample and calculating the RSD for peaks area and retention times.
Recovery studies were carried out to determine the accuracy of SE-HPLC and RP-HPLC methods.
Statistics
The averages of triplicate analysis were calculated for each protein fraction. The results were statistically analysed by analysis of variance. Differences (t-test) were considered significant for P<0·05. Statistical analyses were all performed with SPSS for Windows version 18 (SPSS Inc., Chicago, IL).
Results and Discussion
Effect of thermal processing on qualitative profile of cows' and donkey milk proteins
The results from SE-HPLC analyses of whole cows' and donkey milk (raw and thermally processed) and respective whey (Fig. 1) were compared with those from SDS–PAGE separations (Fig. 2). The behaviour of cows' and donkey milk proteins in Laemmli gels was similar to that observed in earlier reports (Salimei et al. Reference Salimei, Fantuz, Coppola, Chiofalo, Polidori and Varisco2004; Vincenzetti et al. Reference Vincenzetti, Polidori, Mariani, Cammertoni, Fantuz and Vita2008).
Typical chromatographic profile obtained by SE-HPLC for (a) raw whole cows' milk; (b) whey cows' proteins; (c) thermally processed cows' milk; (d) raw donkey milk; (e) whey donkey proteins; (f) thermally processed donkey milk (sample dilution 1:1).
SDS–PAGE of cows' and donkey milk and whey: (1) weight markers; (2) raw whole cows' milk; (3) whey cows' proteins; (4) thermally processed cows' milk; (5) UHT cows' milk; (6) raw donkey milk; (7) whey donkey proteins; (8) thermally processed donkey milk.
Raw cows' milk (Fig. 1a) presented four major peaks corresponding to molecular weight higher than 10 kDa. The first peak corresponds to elution of serum albumin (BSA), lactoferrin (lf), immunoglobulins (Ig) and caseins (with molecular weight higher than 37 kDa), the second peak is from other caseins (molecular weight between 27 and 33 kDa), the third peak is from β-lg (molecular weight around 19 kDa), the fourth peak is from α-la and lys (molecular weight around 14–17 kDa). Additionally, low molecular weight substances (<10 kDa) were observed. Similar chromatographic profile was observed for cows' milk whey (Fig. 1b) except that the peaks from cn were not eluted since these proteins were precipitated at pH 4·6 as observed in SDS–PAGE (Fig. 2).
Chromatographic profile of laboratory thermally processed milk (Fig. 1c) was different from that of raw whole and whey cows' milk. The major chromatographic peaks corresponded to a molecular weight around 66 kDa for the first peak and between 27 and 33 kDa for the second peak, highlighting the occurrence of protein aggregation as a consequence of denaturation. Peaks corresponding to β-lg (molecular weight around 19 kDa), and α-la were not present. Apparently these proteins co-eluted with cn peaks. Analyses of commercial thermal processed samples (UHT treated milk) presented similar chromatographic profile when compared with milk samples that were thermally processed at laboratory.
Donkey milk (Fig. 1d) presented three major peaks corresponding to elution of β-lg, α-la and lys (molecular weight between 19 to 14 kDa). Other small chromatographic peaks were observed. The first small peak corresponds to elution of BSA, lf, and Ig (with molecular weight higher than 66 kDa) and the second and third small peaks are from cn fractions (molecular weight between 27 and 33 kDa). Additionally, low molecular weight substances (<10 kDa) were observed. Similar chromatographic profile was observed for donkey milk whey (Fig. 1e), apparently the sample preparation procedure for HPLC analyses removed part of donkey milk cn and similar chromatographic profiles were obtained for whole milk and whey. The use of frozen samples can justify these results as observed recently by Polidori & Vincenzetti (Reference Polidori and Vincenzetti2010). Chromatographic profile of thermally processed donkey milk (Fig. 1f) showed two major peaks with similar retention time to that of whey proteins. Donkey milk β-lg and α-la suffered different influences of thermal processing when compared with cows' milk β-lg and α-la.
The results from RP-HPLC analyses of whole cows' and donkey milk (raw and thermally processed) and respective whey (Fig. 3) were compared with those from SE-HPLC separations and SDS–PAGE.
Typical chromatographic profile obtained by RP-HPLC for (a) raw whole cow's milk; (b) whey cows' proteins; (c) thermally processed cows' milk; (d) raw donkey milk; (e) whey donkey proteins; (f) thermally processed donkey milk (sample dilution 1:1, except raw whole cows' milk, dilution 1:5).
Raw cows' milk (Fig. 3a) presented four major peaks corresponding to α-la, αs-cn, β-cn and β-lg, respectively. Identification was performed by comparison with cows' milk standards and with literature (Ferreira, Reference Ferreira2005). Traces of lys were observed, identification was performed by comparison with egg lys. Only two major peaks were observed on the chromatographic profile of cows' milk whey (Fig. 3b), corresponding to α-la, and β-lg, respectively. As expected cn fractions were not eluted since these proteins were precipitated at pH 4·6 as observed in SDS–PAGE and SE–HPLC separations. Chromatographic profile of laboratory thermally processed milk (Fig. 3c) and UHT milks was similar, but it was different from that of raw cows' milk and whey. It presented two major chromatographic peaks corresponding to cn fractions. During heat treatment whey proteins unfold and polymerise via disulphide bonds forming aggregates with different hydrophobicity (Curda et al. Reference Curda, Belhácova, Uhrova, Stetina and Fukal1997). Peaks corresponding to β-lg and α-la were not present, since whey proteins denaturation influenced their polarity and consequently they are not analysed by RP-HPLC using the proposed chromatographic conditions (Ferreira et al. Reference Ferreira, Mendes and Ferreira2001). SE-HPLC confirmed that these denatured proteins co-elute with β-cn. The lack of tertiary structure accounts for the stability of cn against heat denaturation because there is very little structure to unfold (Walstra, Reference Walstra1990).
Raw donkey milk (Fig. 3d) showed four major peaks corresponding to elution of lys, α-la, cn and β-lg, respectively. As expected cn peak was not observed in donkey milk whey (Fig. 3e). Chromatographic profile of laboratory processed donkey milk (Fig. 3f) showed two major peaks, no lys was observed and cn peak co-elutes with β-lg. These results indicate that major donkey milk proteins are not strongly affected by thermal processing. According to Polidori & Vincenzetti (Reference Polidori and Vincenzetti2010), donkey milk lys thermal denaturation starts at 70°C, which justifies the absence of lys peak in RP-HPLC chromatogram. However, the effects of thermal denaturation on donkey milk α-la and β-lg are not described in literature.
Effect of thermal processing on quantitative profile of cows' and donkey milk proteins
Lys, β-lg, cn and α-la contents of raw, thermally processed and whey milk samples were determined by SE-HPLC and RP-HPLC. The external standard method was used to calibrate the SE-HPLC and RP-HPLC chromatographic systems. Parameters of calibration curves are reported in Table 1. Linearity between the concentration of β-lg, α-la, lys and cn and the UV absorbance was maintained over the concentration ranges of 1·0–8·0 mg/ml for whole cn and 0·25–2·0 for the other proteins. LOD was higher for cn (0·2 mg/ml) than for other proteins (<0·10 mg/ml). For all peaks, linearity was observed between the amount of protein and the detector response as indicated by r values that exceeded 0·99.
Parameters of regression equations for calibration curves and limit of detection (LOD) for β-lactoglobulin, α-lactoalbumin, lysozyme, casein
† Five points were considered for the regression, except for total casein (cn) (four points were considered). Each point represents the average of three injections of each standard solution
‡ sd in the slope and intercept of the regression line are given in parentheses
§ Correlation coefficient
Values of RSD for retention times and peak areas obtained in the analysis of repeatability and reproducibility are presented in Table 2. Repeatability was evaluated by five consecutive injections of a raw milk sample. The RSD values for retention times (RT) were below 0·64% within analytical day (repeatability) and below 2·10% across analytical days (reproducibility). Values of RSD for peak areas were below 3·32% within day and below 5·64% across days. These results indicate that the precision (repeatability and reproducibility) of the two methods was good and similar to those reported in literature for within- and between-days variation (Bonfatti et al. Reference Bonfatti, Grigoletto, Cecchinato, Gallo and Carnier2008).
Repeatability and reproducibility of SE-HPLC and RP-HPLC, expressed as the relative standard deviation (RSD), and recovery assays in a raw milk sample
Recovery studies were carried out to determine the accuracy of the method (Table 2). The two methods achieved recoveries close to 100% for all proteins. Results of Student's t-test indicated that recovery rates were not significantly different from 100% at P<0·05, except for β-lg quantified by RP-HPLC. Table 3 presents the content of total proteins in cows' and donkey milk including raw milk, whey and thermally processed milk determined by Bradford method. In general, the total protein content was in agreement with literature for cows' and donkey milks and respective whey. Thermally processed cows' and donkey milks presented slightly lower protein content when compared with raw milk. These results are in good agreement with a study concerning the differences of protein fractions among fresh, frozen and powdered donkey milk (Polidori & Vincenzetti, Reference Polidori and Vincenzetti2010).
Total protein content of cows' and donkey whole raw milk, whey and thermally processed milk quantified by Bradford method, including. SE-HPLC and RP-HPLC quantification of β-lactoglobulin, α-lactoalbumin, lysozyme and casein. Mean results of triplicate analysis expressed as mg/ml±sd
† Prepared at laboratory by heating 5 min at 100°C. n.d. not detectable
Concerning cows' milk quantification of β-lg, α-la and cn, no significant differences between results obtained by SE-HPLC and by RP-HPLC were observed for raw milk and whey (paired t-test was used, since the data showed homogeneity of variance and the same samples were analysed by both methods, P>0·05) (Table 3), whereas for thermally processed milk, SE-HPLC indicates an aggregation of proteins presenting high molecular weight (Fig. 1c) and RP-HPLC enables higher cn quantification than in whole milk, probably owing to co-elution of denatured proteins (Fig. 3c). Only traces of lys were observed by RP-HPLC.
With respect to donkey milk quantification of β-lg, α-la, lys, and cn, no significant differences between results obtained by SE-HPLC and by RP-HPLC were observed for raw milk and whey (Table 3) (paired t-test was used, since the data showed homogeneity of variance and the same samples were analysed by both methods, P>0·05). It should be highlight that cn content in donkey milk was low (around 1·4 mg/ml). According to literature higher cn content was expected, however Polidori & Vincenzetti (Reference Polidori and Vincenzetti2010) found substantial differences in cn content, with a value of 3-fold and 2-fold lower in frozen milk with respect to the fresh donkey milk. The raw milk samples used in this study were frozen and centrifugation of donkey milk samples, necessary for HPLC analysis lead to a pellet of cn (confirmed by SE-HPLC analysis, results not shown). Lys was quantified in whole raw milk and whey by SE-HPLC and by RP-HPLC (mean value 1 mg/ml), however thermally processed milk did not present lys peak as a consequence of denaturation. Raw donkey's milk showed higher level of lys compared with raw bovine milk. The amount of lys in raw donkey milk is confirmed by other authors (Vincenzetti et al. Reference Vincenzetti, Polidori, Mariani, Cammertoni, Fantuz and Vita2008; Polidori & Vincenzetti, Reference Polidori and Vincenzetti2010; D'Alessandro et al. Reference D'Alessandro, Martemucci, Jirillo and Leo2011). α-La from donkey milk was stable to thermal processing and concerning β-lg more than 75% of this protein remained stable after 5 min heating at 100°C.
Studies about stability of donkey whey proteins are scarce, it is described that the total whey protein content in powdered donkey milk (quantified by Bradford method) is about 93% of the total whey protein content in donkey frozen milk (Polidori & Vincenzetti, Reference Polidori and Vincenzetti2010). No information was found concerning the thermal stability of β-lg and α-la.
In conclusion, the two chromatographic approaches, SE-HPLC and RP-HPLC gave complementary information to study the effects thermal processing on qualitative and quantitative profile of milk proteins and can be used in routine control of thermal processing of cows' and donkey milk β-lg, α-la, lys and cn. The first chromatographic method highlighted thermal processing changes such as protein aggregation and the second method shows the occurrence of protein denaturation by monitoring peak disappearing owing to changes of protein polarity.
