DETERMINING THE MILK CONTENT OF MILKBASED FOOD PRODUCTS. FSA Final Report Q01117. Campden BRI Project 98164. John J. Dooley, Piotr Jasionowicz, Brian Burch, Sophie Wellum & Helen M. Brown.

Executive Summary The Food Standards Agency (FSA) has recently been focussing on methods that utilise lab-on-a-chip capillary electrophoresis, which is a low-cost, easy-to-use technique that enables profiling of analytes including protein and DNA. Such systems, for example the Agilent 2100 Bioanalyzer, are ideally suited to authenticity applications and have already been adopted by local government laboratories for DNA-based analysis of food. Additional methods that can be applied using this platform will allow these laboratories to widen their testing remit and utilise the full potential of lab-on-a-chip analysis. As capillary electrophoresis techniques have been applied successfully for the quantitative determination of milk proteins, the aim of the current work was to assess the feasibility of transferring capillary electrophoresis-based methods to the simple lab-on-chip platform to assess whether they could be used to determine whether composite food products contain 50% or more of processed animal product, using milk as an example. A protein profiling technique was evaluated for determining milk protein in food products using the Agilent 2100 Bioanalyzer with the Protein 80 LabChip. Protein separation on the 2100 Bioanalyzer can be achieved using one of two protein LabChips, the Protein 80 or the Protein 230. For milk proteins, whilst profiles could be generated on the Protein 230 LabChip, better resolution and sensitivity was achieved with the Protein 80 LabChip and there was less interference from system peaks on this LabChip.

i

Proteins were extracted under reducing conditions using a total protein solubilisation (TPS) buffer.

Alternative extraction methods were investigated but these caused

interference with protein separation on the LabChip. Quantification of 1000µg/ml solutions of milk proteins prepared from commercial αcasein, β-casein, κ-casein, α-lactalbumin and β-lactoglobulin gave recoveries of 103113% when calibrated against the equivalent protein standard curve. For quantification of the amount of milk in milk-based products, β-casein (100-1500µg/ml) and αlactalbumin (50-300µg/ml) were selected and used in a mixed standard solution (containing both proteins) to produce standard curves. Best results were achieved when electropherograms were analysed manually rather than using the auto-analyse function of the 2100 Bioanalyzer software. Manual peak analysis was required to correctly identify some peaks or to define peaks which were not fully resolved. When the method was applied to relatively complex mixtures of proteins (milk proteins mixed with wheat flour or egg powder), milk proteins were not resolved from other proteins, indicating that the method is only likely to be suitable for simple milk-based products. Therefore method validation studies were performed using rice puddings and body building powder (weight gain formula). These were produced at Campden BRI so that the milk content was known on the basis of known weights of ingredients and by compositional analysis of ingredients. These were compared with results generated using the Bioanalyzer method. Rice puddings containing 75%, 50% and 25% milk were produced using whole milk, semiskimmed milk, skimmed milk and Lactofree semi-skimmed milk. The milk content calculated using compositional analysis and recipe data agreed with the actual milk content in all cases except for rice puddings containing 75% whole milk, 75% Lactofree semi-skimmed milk and 25% Lactofree semi-skimmed milk, where the calculated results were 72%, 73% and 26%, respectively.

The milk content determined using the

Bioanalyzer method was very variable (relative standard deviation of up to 60%) and in all cases underestimated the milk content - determining between 30% and 80% of the actual milk content. The lack of precision and accuracy was due to the unpredictable performance of the upper marker used as an internal standard for quantification of

ii

proteins on the LabChips and to the poor resolution of the β-casein peak from α-casein. Rice puddings were sampled by taking liquor only to avoid including rice proteins - this may also have contributed to the lack of accuracy but was anticipated to result in overestimation of milk content rather than underestimation. For the body building powder, the protein profile generated using the Bioanalyzer showed the presence of whey proteins α-lactalbumin and β-lactoglobulin and only traces of casein, showing that for this sample it was more relevant to quantify whey protein than milk content. Quantification using the Bioanalyzer method was based on using α-lactalbumin as the calibrant.

The high α-lactalbumin content of the body

building powder required that the sample be tested using 2mg body building powder/ml solubilisation buffer to ensure that the protein concentration fell within the range of the standard curve appropriate for the Bioanalyzer. The measured α-lactalbumin content and total whey protein content calculated from it indicated that this product contained less than 50% whey protein. However, on the basis of the known weight of ingredients and their composition, the body building powder contained 74.25% whey protein and, on the basis of compositional analysis, 74.9% protein. The underestimate when using the LabChip method may be due to a lack of linearity at the high whey protein concentrations in the body building powder and/or due to the proportion of αlactalbumin in whey isolates and concentrates differing from that found in milk, due to separation and manufacturing technologies employed in the preparation of whey proteins. In conclusion, the validation study showed that the lack of precision and accuracy of the method in its current form makes it unfit for estimating whether the milk content of even simple milk-based products exceeds 50%. For very simple milk protein based products, it may be possible to use the current LabChip technology as a relatively simple, rapid screening method to indicate the presence of a biased peak ratio, i.e. very large αlactalbumin peak compared to β-casein peak, for detecting samples that have been bulked with added whey protein or other milk protein(s), but the LabChip technology in its current form is unsuitable for estimating the milk content of unknown samples to determine whether they contain 50% or more milk.

iii

TABLE OF CONTENTS TABLE OF CONTENTS ............................................................................................................................... IV GLOSSARY ................................................................................................................................................ V ABBREVIATIONS ...................................................................................................................................... VI 1.

INTRODUCTION ............................................................................................................................... 1

2.

MATERIALS AND METHODS ............................................................................................................. 2

2.1.

CHEMICALS AND REAGENTS........................................................................................................... 2

2.2.

PREPARATION OF SAMPLES ........................................................................................................... 2

2.3

PROTEIN EXTRACTION .................................................................................................................. 3

2.4

PREPARATION OF MILK PROTEIN STANDARDS ................................................................................... 3 2.4.1 Preparation of milk protein stock solutions ................................................................................. 3 2.4.2 Preparation of F-milk protein standards ...................................................................................... 5 2.4.3 Preparation of T-milk protein standards ...................................................................................... 5

2.5

ANALYSIS OF SAMPLES TO DETERMINE MILK CONTENT ON THE 2100 BIOANALYZER ............................... 6 2.5.1 Sample denaturation and dilution ............................................................................................... 6 2.5.2 Separation of protein on the LabChip and data analysis ............................................................. 7 2.5.3 Calculation of the milk content .................................................................................................... 8

2.6 3 3.1.

COMPOSITIONAL ANALYSIS ........................................................................................................... 8 RESULTS AND DISCUSSION ............................................................................................................... 9 METHOD DEVELOPMENT .............................................................................................................. 9 3.1.1. Modified De Jong extraction buffer ........................................................................................... 10 3.1.2. Total protein solubilisation buffer (TPS buffer) .......................................................................... 10 3.1.3 BMR extraction buffer ................................................................................................................ 11 3.1.4 Protein molecular weights ......................................................................................................... 12

3.2.

QUANTIFICATION OF MILK PROTEINS ............................................................................................ 13 3.2.1 Composition of milk and milk protein ........................................................................................ 13 3.2.2 Quantification of milk proteins using the Bioanalyzer ............................................................... 14

3.3.

ANALYSIS OF MILK..................................................................................................................... 16

3.4

IN-HOUSE VALIDATION USING MILK PRODUCTS PREPARED AT CAMPDEN BRI ...................................... 17 3.4.1. Production of rice puddings ....................................................................................................... 17 3.4.2. Calculation of milk content from compositional analysis of rice puddings ................................ 18 3.4.3. LabChip analysis of rice puddings and calculation of milk content ............................................ 20 3.4.4. LabChip analysis of body building powder ................................................................................. 25

4

CONCLUSIONS................................................................................................................................ 27

iv

5

AREAS FOR FUTURE INVESTIGATION.............................................................................................. 30

6

ACKNOWLEDGEMENTS .................................................................................................................. 30

7

REFERENCES ................................................................................................................................... 31

8

APPENDICES ................................................................................................................................... 33

APPENDIX 8.1 PREPARATION OF SAMPLES AT CAMPDEN BRI ................................................................................... 33 8.1.1 Preparation of canned rice puddings .......................................................................................... 33 8.1.2 Preparation of a body building powder ..................................................................................... 34 APPENDIX 8.2 TYPICAL RESULTS OBTAINED USING THE PROTEIN 80 LABCHIP .............................................................. 36 APPENDIX 8.3 EXAMPLE CALCULATION OF THE MILK CONTENT OF RICE PUDDING ....................................................... 40 APPENDIX 8.4 CALCULATION OF WHEY PROTEIN IN BODY BUILDING POWDER ............................................................ 41 APPENDIX 8.5 STANDARD OPERATING PROCEDURE ............................................................................................... 42 9.

FIGURES ......................................................................................................................................... 64

GLOSSARY 2100 Bioanalyzer: A small-scale capillary electrophoretic system using lab-on-a-chip technology and microfluidics for the specific separation of DNA fragments or proteins. Capillary electrophoresis (CE): Electrophoresis in narrow bore capillaries, normally 25100µm internal diameter. Electrophoresis may be in free solution, i.e. capillary zone electrophoresis (CZE); in a sieving matrix, i.e. capillary gel electrophoresis (CGE); or in a partitioning matrix, i.e. micellar electrokinetic chromatography (MEKC). Fluorescent Units (FU): A measure of fluorescence intensity used by the 2100 Bioanalyzer. Electrophoresis: Differential movement or migration of ions by attraction or repulsion in an electric field. Gel electrophoresis: Electrophoresis performed in a gel of acrylamide or agar. Proteins migrate through the gel when an electric current is applied. The gel matrix acts as a sieve to separate the proteins based on size and charge.

v

Protein LabChip: A small (3cm2), disposable, single-use plastic and glass unit containing etched capillaries attached directly to ten sample loading wells. Two protein LabChips are available, the Protein 80 LabChip for separating fragments of 5-80kDa and the Protein 230 LabChip for separating 14-230kDa fragments.

ABBREVIATIONS AA - amino acid α-Lg - α-lactoglobulin β-Lg - β-lactoglobulin BBP - body building powder BMR - a solubilisation buffer prepared from phosphate buffered saline (PBS) buffer, pH7.4, containing 8M urea CE - capillary electrophoresis CZE - capillary zone electrophoresis DL-DTT

- dithiothreitol

FSA - Food Standards Agency PBS - phosphate buffered saline QUID - quantitative declaration of ingredients SDS - sodium dodecyl sulphate TPS buffer - total protein solubilisation buffer WPC - whey protein concentrate WPI - whey protein isolate

vi

1. INTRODUCTION The ability to quantify the level of ingredients in composite food products provides a means of monitoring products for consumer protection and regulatory compliance. At the time of commissioning this work, changes to the European Commission Regulation covering veterinary checks on imported products of animal origin were expected. It was anticipated that they would stipulate checks to be carried out on composite foods imported from third countries that contain 50% or more processed animal product (other than meat, for which separate controls exist) by weight of the food. The FSA, in its role of protecting consumers by effective enforcement of food legislation, identified the need for reliable methods to identify foods that would require veterinary checks, as there is no existing simple, rapid test to determine the level of animal-based ingredients in composite foods. The Food Standards Agency (FSA) has supported the development of easy-to-use approaches for analysing and quantifying food ingredients.

Lab-on-a-chip capillary

electrophoresis (CE), using the Agilent 2100 Bioanalyzer, offers considerable advantages over standard gel electrophoresis for protein analysis. The Agilent 2100 Bioanalyzer is available with a choice of two Protein LabChips, the Protein 80 LabChip or the Protein 230 LabChip. The Protein 80 LabChip is designed for sizing and analysis of proteins from 5-80kDa and the Protein 230 LabChip is designed for sizing and analysis of proteins from 14-230kDa. The relatively low cost and ease of use of the Bioanalyzer for DNA analysis has led to the adoption of this system by a number of government enforcement and private laboratories. The aim of the current study was to assess the feasibility of transferring existing CE-based methods for the quantitative determination of milk proteins to this simple lab-on-a-chip, CE platform and to assess whether it could be used to determine whether composite food products contain 50% or more of processed animal product. Milk was chosen as an example of an animal-based ingredient to assess the feasibility of this approach. Quantitative analysis of specific analytes in food products relies upon obtaining reliable standards against which the analyte in the unknowns can be measured. When the analyte is 'milk' this is complicated by: the number of different types of milk available for 1

use in food products (e.g. whole milk, skimmed milk, dried milk, whey powder), the type of processing applied to the milk (e.g. UHT, pasteurised), and variation in the composition of milk that occurs due to other factors such as cattle genotype, season or lactation[1-5]. Milk is an aqueous solution of casein (αs-casein, β-casein, κ-casein, γcasein) and whey proteins (lactalbumin, lactoglobulin, immunoglobulins).

The

proportions of the different proteins can vary considerably. By using CE to generate a profile of milk proteins, it was envisaged that the profile would provide an initial screen to identify significant variations in milk protein composition, for example inclusion of whey protein, and allow for quantification of 'milk' on the basis of one or more appropriate proteins. In this work, the method was initially optimised and then validated in-house using composite food products containing 25%, 50% and 75% milk prepared at Campden BRI. These products were analysed using the CE method and traditional methods of compositional analysis. This allowed the milk content determined from quantities of ingredients, compositional analysis and CE analysis to be compared.

2. MATERIALS AND METHODS 2.1.

CHEMICALS AND REAGENTS

All chemicals were obtained from Sigma-Aldrich (Poole, Dorset, UK), unless otherwise stated. Protein profiles were generated using Protein LabChips and specific reagents from Protein 80 (P/N 5067-1515) or Protein 230 (P/N 5067-1517) kits and the Agilent 2100 Bioanalyzer (Agilent Technologies UK Ltd, Stockport, Cheshire, UK).

2.2.

PREPARATION OF SAMPLES

Samples were prepared in-house at Campden BRI (Table 1) using pilot scale equivalents to normal commercial processing equipment (Appendix 8.1) or were purchased from local retailers.

2

Table 1: Samples prepared at Campden BRI. Sample Description Canned rice pudding made with 75%, 50% or 25% whole milk Canned rice pudding made with 75%, 50% or 25% semi-skimmed milk Canned rice pudding made with 75%, 50% or 25% skimmed milk Canned rice pudding made with 75%, 50% or 25% Lactofree semi-skimmed milk Body building powder (high protein weight gain powder)

2.3

PROTEIN EXTRACTION

Prior to extraction, samples were stirred or shaken thoroughly to ensure sample homogeneity. In order that proteins were reduced and that protein denaturation took place under reducing conditions, samples were extracted in a total protein solubilisation buffer (TPS buffer) that contained DL-dithiothreitol (DL-DTT). All samples and standards were extracted using a total protein extraction protocol. Samples (up to 1,000mg) were weighed into clean glass bijou bottles. The exact weight of each sample was noted. The minimum volume of TPS buffer (2M urea, 15 % glycerol, 0.1 M DL-DTT and 0.1 M Tris-HCl, pH 8.8) required to solubilise the sample was added to the bijou bottle. Liquid samples were mixed thoroughly with TPS buffer. Solid samples were dissolved in TPS buffer by gentle shaking or agitating the sample using a Spiramix5 mixer (Thermo Fisher Scientific). The samples were then centrifuged in an Eppendorf centrifuge (5415D) at 2,000g for 15 minutes to remove particulate material from the supernatant. Aliquots (4µl) of the supernatant were denatured prior to analysis on the 2100 Bioanalyzer.

2.4

PREPARATION OF MILK PROTEIN STANDARDS 2.4.1 PREPARATION OF MILK PROTEIN STOCK SOLUTIONS

Individual milk protein powders (Table 2) were weighed into separate glass bijou bottles. The exact weight of protein taken was noted. Exact volumes of TPS buffer were added to achieve milk protein stock solutions of 1,000µg/ml, 5,000µg/ml or 10,000µg/ml (or 1, 5 or 10mg/ml, respectively) (Table 3). Bijou bottles of stock solution were placed on a 3

Spiramix5 mixer for 1 hour to ensure that the proteins were fully hydrated, well mixed and homogeneous. Table 2: Milk proteins evaluated for use as standards. Milk Protein

Purity by PAGE

Source

β-lactoglobulin

94% as b-Lg A and b-Lg B

Sigma L0130 Lot 095K7006

α-lactalbumin, from bovine milk Type III, calcium depleted.

98%

Sigma L6010 Lot 035K7005

α-casein, from bovine milk >70% as as-casein

90%

Sigma C6780 Lot 075K7425

β-casein, from bovine milk >90% as beta-casein

99%

Sigma C6905 Lot 105K7410

κ-casein

Minimum 70% κ-casein

Sigma C0406 Lot 051K7575

After 1 hour, 500µl of the 5,000µg/ml β-casein solution was mixed with 750µl of TPS buffer in a clean glass bijou bottle to achieve a 2,000µg/ml stock solution of β-casein. The 2,000µg/ml β-casein stock solution was mixed on a Spiramix5 mixer for 30 minutes to achieve solution homogeneity. Table 3: Preparation of milk protein stock solutions. Milk protein

Approximate weight taken (mg)¹

Approximate volume of TPS buffer (ml)²

Protein stock concentration (µg/ml)

Protein stock concentration (mg/ml)

α-casein

20

2

10,000

10

β-casein

20

2

10,000

10

β-casein

10

2

5,000

5

κ-casein

20

2

10,000

10

α-lactalbumin

20

2

10,000

10

α-lactalbumin

5

5

1,000

1

β-lactoglobulin

20

2

10,000

10

¹ The exact weight of protein taken was noted. ² The actual volume of buffer added was dependent on the actual weight of protein taken and the desired final concentration.

4

2.4.2 PREPARATION OF F-MILK PROTEIN STANDARDS A milk protein standard, referred to as F-milk protein standard, containing each of the five milk proteins at 2,000µg/ml, was prepared from the 10,000µg/ml milk protein stock solutions. Aliquots (400µl) of the individual milk protein stock solutions were mixed in a clean glass bijou bottle. The 2,000µg/ml F-milk protein standard solution was mixed on a Spiramix5 for 30-60 minutes to ensure that solution homogeneity was achieved. A series of F-milk protein standards at 1,500; 1,000; 500 and 250µg/ml was prepared by diluting the 2,000µg/ml F-milk protein standard solution using TPS buffer as detailed in Table 4. The F-milk protein standards were mixed on a Spiramix5 mixer for 30-60 minutes to ensure homogeneity Table 4: Preparation of the F-milk protein standards. Final concentration of F-milk protein standard (µg/ml)

Volume of 2,000µg/ml total milk protein standard (µl)

Volume of TPS buffer (µl)

Final volume (µl)

2,000

375

~

375

1,500

750

250

1,000

1,000

500

500

1,000

500

250

750

1,000

250

125

875

1,000

2.4.3 PREPARATION OF T-MILK PROTEIN STANDARDS A series of milk protein standards, referred to as T-milk protein standards, was prepared in bijou bottles, from the α-lactalbumin and β-casein stock solutions (Table 5). Where necessary, T-milk protein standards were diluted using TPS buffer.

T-milk protein

standards were mixed on a Spiramix5 mixer for 30-60 minutes to ensure that the standards were homogeneous. The final volume and concentration of each milk protein in the T-milk protein standards are shown in Table 6.

5

Table 5: Volumes of stock milk protein solution used to prepare the T-milk protein standards. T-milk protein standard

Volume of each stock solution (µl)

Final volume (µl)

5,000µg/ml βcasein stock

2,000µg/ml βcasein stock

1,000µg/ml αlactalbumin stock

TPS buffer

Standard 1

300

-

300

400

1000

Standard 2

-

500

200

300

1000

Standard 3

-

250

150

600

1000

Standard 4

-

125

100

775

1000

Standard 5

-

50

50

900

1000

Table 6: Final concentration of α-lactalbumin and β-casein proteins in the T-milk protein standards.

2.5

T-milk protein standard

Final concentration of β-casein (µg/ml)

Final concentration of α-lactalbumin (µg/ml)

Standard 1

1,500

300

Standard 2

1,000

200

Standard 3

500

150

Standard 4

250

100

Standard 5

100

50

ANALYSIS OF SAMPLES BIOANALYZER

TO

DETERMINE MILK CONTENT

ON THE

2100

2.5.1 SAMPLE DENATURATION AND DILUTION All samples and protein standards, used to generate standard curves, were denatured using denaturing solution supplied in the Agilent Protein 80 reagent kit. An aliquot (4µl) of the sample or standard was combined with 2μl of denaturing solution in a 0.5ml microtube. The lids were closed and secured, using a tube cap-lock, to prevent them opening during heating. The tubes were boiled at 95°C for 5 minutes using a water bath. In addition to the samples, a 6μl aliquot of the Protein 80 ladder was heated in a 0.5ml microtube at 95°C for 5 minutes. The ladder was heated with the first batch of samples and was used for the analysis of all LabChips analysed on the day of heating.

6

After boiling for 5 minutes, the tubes were left to cool to room temperature before they were centrifuged at 16,000g for 15 seconds to recover the solution in the bottom of the tube. The boiled solutions (samples, standards and ladder) were mixed with 84μl of ultrapure water, by vortexing, before the tubes were centrifuged at 16,000g for 15 seconds to recover the solution ready for loading onto the Protein 80 LabChip. All sample extracts, stock solutions, standards and the ladder were used within one day and were stored on ice when not in use.

2.5.2 SEPARATION OF PROTEIN ON THE LABCHIP AND DATA ANALYSIS Protein profiles were generated on the 2100 Bioanalyzer according to the instructions in the Agilent Protein 80 Kit Guide. Denatured sample solution or ladder (6µl) was added to sample or ladder wells on the LabChip. CE was started within two minutes of preparing the LabChip. Separation conditions are fixed and pre-determined by the Protein chip assay for the Bioanalyzer. Data analysis was performed using the 'Protein 80 Series II Assay' of the Agilent 2100 Expert (Version B.02.03) software. The 2100 Expert software was run using the absolute quantification method with manual integration to ensure that protein peaks were detected and correctly assigned. The concentration of either α-lactalbumin or β-casein was determined using the appropriate standard curve. The Agilent software does not have a facility to allow data analysis to be performed using two standard curves simultaneously. In order to perform analysis of the sample data, it was therefore necessary to complete the analysis using the α-lactalbumin protein standards and then repeat the analysis using the β-casein protein standards. This required resetting all the values for the protein standard concentrations and in some cases manually selecting the correct protein peaks in the profiles. In addition, during the analysis of some samples both the α-casein and β-casein peaks were not fully resolved. Despite using manual integration to define the two protein peaks, precise measurement of the β-casein peak was compromised, making a significant contribution to the overall uncertainty of the analysis.

7

2.5.3 CALCULATION OF THE MILK CONTENT The 2100 Bioanalyzer was used to determine the concentration of specific milk protein, either α-lactalbumin or β-casein, in the extracted solubilised sample applied to the LabChip. The following calculations were used to convert this into "milk content": The concentration (mg/ml) of milk protein (α-lactalbumin or β-casein) in the sample (A) was calculated using the following formula: A

P DF 1000

where 'A' is the concentration of either α-lactalbumin or β-casein in the sample 'P' is the concentration of protein (α-lactalbumin or β-casein) in the sample going onto the LabChip (µg/ml) and 'DF' is the dilution factor prior to denaturation, i.e. at extraction and solubilisation. The percentage "milk content" of the sample was calculated: A 100% CF

where 'A' is the concentration of either α-lactalbumin or β-casein in the sample 'CF' is a factor derived from the concentration of either α-lactalbumin or β-casein in 100% whole milk, i.e. an 'α-lactalbumin factor' or a ' β-casein factor'. (Note this is the same principle as a nitrogen factor used for determining the protein content from analysis of organic nitrogen in proteins).

2.6

COMPOSITIONAL ANALYSIS

Compositional analysis was performed at Campden BRI on the rice puddings, the body building powder and the ingredients used for their preparation. All methods used are

8

accredited by the United Kingdom Accreditation Service (UKAS) and are summarised in Table 7. Table 7: Methods applied to determine the compositional analysis of ingredients and products used in this study. Method code

Method name

TES-AC-086

Determination of Total Ash in Foodstuffs

TES-AC-087

Determination of Organic Nitrogen (and Protein by Calculation) in Foodstuffs by Automated 'Kjeltec' System

TES-AC-097

Determination of Moisture in Foodstuffs

TES-AC-202

Determination of Fat in Foodstuffs - Rose Gottlieb Extraction

TES-AC-335

Method for the Calculation of the Energy Content of Foods from Nutritional Data

TES-AC-536

Determination of Total Fat Content by Weibull Stoldt (Acid Hydrolysis followed by Soxtec Extraction) and Crude Fat (Soxtec Extraction)

TES-AC-628

Determination of Moisture and Fat Content in Foods using the CEM Smart Trac System

3 RESULTS AND DISCUSSION 3.1

METHOD DEVELOPMENT

In order to determine the optimal protein solubilisation buffer for use with milk products, three different extraction buffers were investigated. Investigations were initially performed using the purified milk proteins α-lactalbumin, β-lactoglobulin, αcasein, β-casein and κ-casein. After solubilisation, proteins were separated on the Agilent 2100 Bioanalyzer using two different protein LabChips, the Protein 230 LabChip and the Protein 80 LabChip, which resolve proteins of 14-230kDa or 5-80kDa, respectively. The expected protein sizes (Table 8) indicated that it should be possible to resolve the proteins on either LabChip; however, protein separation was initially performed using both LabChips to identify the optimal LabChip for milk protein quantification.

9

Table 8: Comparison of milk protein molecular weight determined from amino acid (AA) composition and the 2100 Bioanalyzer with Protein 80 LabChip. Protein

Molecular weight of proteins determined using AA composition¹ [kDa]

2100 Bioanalyzer [kDa]

α-lactalbumin

14.1

12.2

β-lactoglobulin

18.3

18.3

αs1-casein

23.6

37.7²

αs2-casein

22.5

β-casein

24

32.4

19

45.1

κ-casein [9]

¹ Data from reference Miralles et al 2000 ² The two α-casein sub-units (αs1 & αs2) were not differentiated on the 2100 Bioanalyzer

3.1.1. MODIFIED DE JONG EXTRACTION BUFFER Milk protein standards were analysed using a solubilisation buffer that was based on a protocol that had been developed previously to solubilise milk proteins for analysis by capillary zone electrophoresis (CZE)[6]. This buffer was, therefore, considered potentially useful for solubilising proteins for analysis on the 2100 Bioanalyzer. Individual milk protein standards (20mg) were dissolved in 10ml of ‘De Jong’ buffer (6M Urea, 5mM DLdithiothreitol (DL-DTT), 5mM disodium phosphate, pH 8.0) to achieve a final concentration of 2000µg/ml (or 2mg/ml). Equal volumes of the five standards were mixed and, where necessary, diluted with the buffer to produce mixed standards at 400, 300, 100, 50 and 20µg/ml. Analysis of samples extracted with the De Jong buffer resulted in poor resolution of the protein profiles with both the Protein 230 and Protein 80 LabChips (Figure 1).

Results of analysis with either LabChip produced

electropherograms with lost upper markers, high backgrounds or only a few broad peaks, suggesting that the De Jong buffer was interfering with the separation of proteins on the 2100 Bioanalyzer. It was, therefore, concluded that this buffer was not suited to the extraction and analysis of milk proteins using the Agilent 2100 Bioanalyzer.

3.1.2. TOTAL PROTEIN SOLUBILISATION BUFFER (TPS BUFFER) Milk protein standards were solubilised using TPS buffer. This contained 0.1M DTT to ensure that protein extraction took place under reducing conditions. Protein profiles

10

were produced using both Protein 80 and Protein 230 LabChips (Figure 2). Profiles generated on the Protein 80 LabChip showed that good separation of all five milk proteins was achieved and that all five proteins were detected even when loaded at 250µg/ml. When proteins (at a concentration of 2000µg/ml) were profiled on the Protein 230 LabChip, only the α-lactalbumin, β-lactoglobulin and κ-casein peaks were resolved. The α-casein and β-casein proteins were not resolved on this LabChip. When the proteins were run at a concentration of 250µg/ml, the α-lactalbumin peak was lost and only small peaks corresponding to the β-lactoglobulin, κ-casein and the combined αcasein and β-casein peak were observed. Profiles from both LabChips showed low background noise. System peaks were observed in all profiles generated on the Protein 230 LabChips, but only when low concentrations of protein were analysed on the Protein 80 LabChip. The relative height of any observed system peaks increased as the amount of protein loaded was reduced; however, it was possible to manually exclude system peaks from the analysis to avoid erroneous calculations. Poor detection of the milk proteins and failure to resolve α-casein and β-casein when using the Protein 230 LabChip were considered to indicate that this LabChip was less suitable for the analysis of milk proteins compared to the Protein 80 LabChip. In a previous study of milk proteins using a Protein 200 LabChip, α-lactalbumin co-migrated with one of the LabChip markers, whereas successful separation had been achieved using a Protein 50 LabChip[7]. Both the Protein 50 and Protein 200 LabChips have now been superseded by the Protein 80 and 230 LabChips. Similar protein separation results were achieved using the newer LabChips.

3.1.3 BMR EXTRACTION BUFFER Analysis of the milk proteins was also performed using a solubilisation buffer prepared from phosphate buffered saline (PBS) buffer, pH 7.4, containing 8M urea (BMR buffer)[7]. The PBS solution contained 0.138M NaCl and 2.7mM KCl. Samples were extracted and diluted using the BMR buffer as described for the previous buffers. Separation of samples solubilised in this buffer using either the Protein 80 or Protein 230 LabChips produced variable results (Figure 3). Whilst proteins were separated on both LabChips, better resolution was achieved on the Protein 80 LabChip compared with the Protein 230 LabChip. When higher concentrations (2000µg/ml) of protein were run on the 11

LabChip, higher peaks were produced compared to peaks produced when smaller (250µg/ml) amounts of protein were loaded. However, a greater number of small peaks were also detected when greater amounts of protein were loaded onto the LabChips. These small peaks could be removed from the analysis by changing the baseline threshold level. In contrast, as the amount of protein injected on to the LabChip decreased, an increase in the height of one or more system peaks was observed. The height of these system peaks meant they could not be removed by changing the threshold level, therefore they required manual removal. Overall, the results obtained using BMR buffer were much better than those produced using the modified De Jong buffer.

When results obtained using the BMR extraction buffer (Figure 3) were

compared with those obtained using the total TPS buffer (Figure 2) a number of differences were observed. Resolution of the α-casein and β-casein peaks was not as clear with the BMR buffer as was achieved using the TPS buffer. In addition, the whey proteins both appear to be suppressed when using the BMR buffer compared to the TPS buffer. For example, the β-lactoglobulin peak appears to be greater than (or at least equal to) the α-casein and β-casein peaks when extracted using the TPS buffer; however, when solubilised with the BMR buffer, the height of the β-lactoglobulin peak was consistently less than that of the casein peaks. This suggests that the BMR buffer may be less efficient at solubilising whey proteins or that a preferential extraction of caseins occurs when using this buffer. All subsequent work was, therefore, performed using the Protein 80 LabChip and the TPS buffer.

3.1.4 PROTEIN MOLECULAR WEIGHTS When using the Protein 80 LabChip in auto-analyse mode, it was observed that there was a difference between the protein molecular weights as determined by the 2100 Bioanalyzer and the expected molecular weights. The molecular weights of the whey proteins were close to those expected, but the sizes of the caseins were higher than expected (Table 8). The difference between the expected and observed size of the κcasein is particularly noticeable, as the size determined on the 2100 Bioanalyzer is over twice the expected size. A difference in protein molecular weight due to the different polymeric sieving matrices and field strengths of sodium dodecyl sulphate 12

polyacrylamide gel electrophoresis (SDS-PAGE) compared to sodium dodecyl sulphate capillary electrophoresis (SDS-CE) has been reported previously[9]. It is also known that the binding capacity of sodium dodecyl sulphate (SDS) differs for the individual caseins and is to some extent dependent on competition, as caseins either bind with other casein molecules or with SDS. These previously reported differences could account for observations made here. In addition, the individual protein sizes did not vary when separated on different LabChips, suggesting that sizing on the Protein 80 LabChip is consistent and therefore unlikely to affect the detection or measurement of protein levels in unknown samples.

3.2

QUANTIFICATION OF MILK PROTEINS

3.2.1 COMPOSITION OF MILK AND MILK PROTEIN Average values for the composition of milk from McCance & Widdowson[10] show that the level of total protein in each of the liquid milk types is similar (3.2-3.4%) but the proportion of fat and water differs (Table 9).

Table 9: Average percentage levels of different components of milk. (McCance & Widdowson [10]). Type of milk product

Water (%)

Protein (%)

Fat (%)

Whole cow’s milk, pasteurised

87.8

3.2

3.9

Whole cow’s milk, UHT

87.8

3.2

3.9

Semi-skimmed cow’s milk, pasteurised

89.8

3.3

1.6

Semi-skimmed cow’s milk, UHT

89.7

3.3

1.7

Skimmed cow’s milk, pasteurised

91.1

3.3

0.1

Skimmed cow’s milk, UHT

91.1

3.4

0.1

Dried whole milk powder

2.9

26.3

26.3

Dried skimmed milk

3

36.1

0.6

Evaporated milk (whole)

69.1

8.4

9.4

Milk protein is composed of a number of different proteins (Figure 4A and Table 10). Table 10 summarises the proteins and their concentration in milk as published in various sources[3,11-15]. Reported levels for each protein differ between publications; however, these differences are relatively small. In general, the protein fraction of whole milk 13

comprises 79-83% casein, most of which is α-casein, and 17-20% whey protein, of which β-lactoglobulin contributes the major proportion. The published concentrations of milk proteins were used to calculate an average value for each protein. The average values were used to calculate a 'factor' to calculate the 'milk' content from the determination of specific proteins.

3.2.2 QUANTIFICATION OF MILK PROTEINS USING THE BIOANALYZER The separation of milk proteins on the 2100 Bioanalyzer was assessed using a series of five mixed standards of α-casein, β-casein, κ-casein, α-lactalbumin and β-lactoglobulin at 250µg/ml to 2000µg/ml. Separation of the five milk proteins was achieved on the LabChip (Figure 2 Protein 80 LabChip). Each protein from the mixed standard was used in turn to produce a standard curve and calculate the concentration of a single 1,000 µg/ml protein solution loaded onto the same LabChip. Results (except those for κcasein) were calculated using the auto-analyse function on the Bioanalyzer software to define the peak areas and heights. The mean of two runs (except for α-lactalbumin) made on two separate LabChips is shown in Table 11. Standard curves showed a near straight line between points with the linear regression line having a fit (R 2) of over 95%. Recovery was between 102 and 113% for α-lactalbumin, β-lactoglobulin, α-casein, βcasein and κ-casein.

Generally the use of the auto-analyse function of the 2100

Bioanalyzer software produced a satisfactory determination of the concentration of protein in the solutions.

Some problems were observed with the automatic

identification of κ-casein and β-lactoglobulin peaks; however, using the manual analysis mode overcame these problems. Results obtained using the single protein solutions indicated that, provided no interference due to the sample occurred, it should be possible to quantify the amount of individual milk proteins in unknown samples.

14

Table 10: Concentrations of specific proteins in milk. Published values are shown along with relevant references. Calculated values (converted to mg protein per ml of whole milk) are shown underlined¹. The average calculated values are shown in bold text in the last row. Total protein

Total casein

Total whey

α–casein

β–casein

κ–casein

α–lactalbumin

β–lactoglobulin

Reference

3.39%

79.60% of milk protein

17.60% of milk protein

16.2mg/ml

8.4mg/ml

2.4mg/ml

1.5mg/ml

4.5mg/ml

[3] – Table 1

3.39g/100ml

27g/L

5.9g/L

16.2mg/ml

8.4mg/ml

2.4mg/ml

1.5mg/ml

4.5mg/ml

-

3.4g/100ml

25g/L

6g/L

54% of total casein

33% of total casein

13% of total casein

19% of total whey protein

49% of total whey protein

[14] – Tables 12.1 & 12.2

3.4g/100ml

25g/L

6g/L

13.5mg/ml

8.3mg/ml

3.3mg/ml

1.1mg/ml

2.9mg/ml

-

nr

nr

nr

12.6g/ L

9.3g/ L

3.3g/ L

1.2g/ L

3.2g/ L

[14] – Table 2.1

-

-

-

12.6mg/ml

9.3mg/ml

3.3mg/ml

1.2mg/ml

3.2mg/ml

-

30-35g/L

~78%

nr

50% of total casein

35% of total casein

12% of total casein

1.2g/L

2.7g/L

[15]

3.3g/100ml

26g/L

-

13mg/ml

9.1mg/ml

3.1mg/ml

1.2mg/ml

2.7mg/ml

-

-

80-83% of milk protein

17-20% of milk protein

39-45% of milk protein

21-28% of milk protein

10-12% of milk protein

2-4% of milk protein

9-10% of milk protein

[11-13]

33.6mg/ml

26mg/ml

6mg/ml

13.8mg/ml

8.8mg/ml

3.0mg/ml

1.3mg/ml

3.3mg/ml

¹ Calculated values were derived from published data as part of this study. nr: no reported data. Missing data were not reported by source.

15

This information, in conjunction with data from the literature about the level of each specific protein in milk (Table 10), could be used to determine the amount of milk in products. Rather than using a standard containing all five proteins to determine the milk content, two proteins, one representing the whey fraction (α-lactalbumin) and one the casein fraction (β-casein) were selected for quantification of milk content. These two proteins correspond to protein peaks that migrate at about 22 and 30 seconds, respectively (Figure 5). β-lactoglobulin and κ-casein were excluded due to their requirement for manual analysis and poorer recovery compared to other proteins. Caseins are heat stable so are considered to be suitable for quantification of milk regardless of heat processing. The whey protein provides for products to which whey protein concentrates or isolates have been added or are the sole source of milk protein. Table 11: Determination of protein concentration of single protein solutions using the auto-analyse function of the 2100 Bioanalyzer software. Protein

Actual concentration (µg/ml)

Size [kDa]

Calculated concentration (µg/ml)

Percent recovery

α-lactalbumin

1000

12.2

1,045¹

105%

β-lactoglobulin

1000

18.3

1,131

113%

α-casein

1000

37.7

1,025

102%

β-casein

1000

32.4

1,046

105%

κ-casein

1000

45.1

1,098²

110%

¹ Result of single measurement ² Result obtained using manual integration

3.3

ANALYSIS OF MILK

Milk analysed using the TPS buffer and the 2100 Bioanalyzer showed resolution of the five milk proteins α-lactalbumin, β-lactoglobulin, α-casein, β-casein and κ-casein (Figure 4B). Different types of milk were analysed. Very little difference in the profiles of dried, whole and skimmed milk and liquid skimmed and semi-skimmed and UHT semi-skimmed milk were observed (Figure 6, shows some, but not all, of the milk profiles in simulated gel format for ease of comparison). All five milk proteins were detected in all the milk types. However, resolution was not as good as that achieved using CZE where α-s1 and α-s2 forms of α-casein and A1 and A2 forms of β-casein were resolved (Figure 4A). On

16

the LabChip, protein separation is based on sieving in a gel matrix so is dependent on molecular weight, whereas in CZE, separation is based on the charge on proteins and to a lesser extent the size. This difference affected the resolution and the order in which proteins migrated. However, when using the 2100 Bioanalyzer migration was consistent and, therefore, did not affect the ability to identify the proteins. The lack of resolving power when using the LabChip compared to CZE did suggest that there may be difficulties when analysing products containing complex mixtures of proteins, so analysis of more complex products, i.e. products that contained more than just milk proteins, was assessed using combinations of dried milk powder and 30% or 60% dried egg powder or wheat flour (Figure 7). In the presence of additional proteins, the lack of resolution when using the LabChip meant that milk proteins could not be clearly identified. In particular, the whey proteins were difficult to detect due to their low level compared to other proteins. In addition, proteins from some samples migrated with the upper marker making detection of this marker difficult.

This interfered with the

quantification of milk proteins as the upper marker is used by the 2100 Bioanalyzer software to quantify the amount of protein in each well. These results suggested that quantification of milk proteins in samples containing high levels of proteins from other ingredients may not be possible and the method is only suited to the determination of milk in relatively simple milk products.

Therefore, validation of the method was

performed using rice puddings and whey powder based body building powders as examples of simple milk products.

3.4

IN-HOUSE VALIDATION USING MILK PRODUCTS PREPARED AT CAMPDEN BRI

3.4.1. PRODUCTION OF RICE PUDDINGS Rice puddings (made at Campden BRI using a recipe based on commercial rice pudding recipes) were made using 1% dried whey powder, 9% rice, 4.6% sugar and whole, skimmed or semi-skimmed milk at either 25%, 50% or 75%. Fresh milk was obtained from local retailers. The total recipe was brought to 100% by the use of an appropriate amount of water. A similar recipe was used to produce lactose-free rice puddings (which were made with Lactofree semi-skimmed milk); however, whey protein isolate was added to these products as the standard dried whey powder contained lactose. Initially two cans of each rice pudding, except the Lactofree pudding, were produced to

17

confirm that products were suitable for analysis on the 2100 Bioanalyzer. A further batch of cans of each rice pudding recipe was produced for analytical measurement; one can was used for compositional analysis and the others were used to determine the milk content using the Bioanalyzer. On opening the cans, differences in the colour of the rice puddings were evident (Figure 8). Some puddings were a pale, slightly off-white colour that is typical of rice pudding, whilst others were a darker brown colour. These differences generally corresponded with the amount and type of milk in the cans, suggesting that the reactions between proteins and non-reducing sugars (Maillard reactions) had occurred to a greater or lesser extent during heat processing.

3.4.2. CALCULATION OF MILK CONTENT FROM COMPOSITIONAL ANALYSIS OF RICE PUDDINGS

Results of the compositional analyses carried out on the rice puddings and individual ingredients are shown in Table 12. The composition of milk used was similar to the average values reported by McCance and Widdowson[10] (Table 9). The Lactofree semiskimmed milk contained a slightly higher proportion of protein, but the removal of lactose was mainly balanced by an increase in water. Compositional analysis of the rice puddings showed that as the percentage of milk in the whole milk rice puddings dropped from 75% to 25%, the protein level dropped from 3.0g/100g to 1.5g/100g, i.e. a 50% reduction. Similar changes were observed in rice puddings made with other milk types, i.e. the protein levels were not affected by the type of milk added. Because the protein concentration of the various forms of milk remains unchanged, it should be feasible to calculate the milk content on the basis of milk protein. Changes in the fat, carbohydrate or total energy levels of the products also reflected reductions in the milk content; however, the actual values measured depended on the type of milk.

A

relatively large (2.9%) change in the carbohydrate level was noted between the whole milk rice puddings, while relatively small (