J. Dairy Sci. 90:602-615
© American Dairy Science Association, 2007.
Lipolysis and Proteolysis of Modified and Producer Milks Used for Calibration of Mid-Infrared Milk Analyzers1
K. E. Kaylegian*,
J. M. Lynch*,
J. R. Fleming
and
D. M. Barbano*,2
* Northeast Dairy Foods Research Center, Department of Food Science, Cornell University, Ithaca, NY
USDA, Agricultural Marketing Service, Southwest Milk Marketing Area, Carrollton, TX
2 Corresponding author: dmb37{at}cornell.edu
 |
ABSTRACT
|
|---|
Our objective was to determine if lipolysis or proteolysis of calibration sets during shelf life influenced the mid-infrared (MIR) readings or calibration slopes and intercepts. The lipolytic and proteolytic deterioration was measured for 3 modified milk and 3 producer milk calibration sets during storage at 4°C. Modified and producer milk sets were used separately to calibrate an optical filter and virtual filter MIR analyzer. The uncorrected readings and slopes and intercepts of the calibration linear regressions for fat B, fat A, protein, and lactose were determined over 28 d for modified milks and 15 d for producer milks. It was expected that increases in free fatty acid content and decreases in the casein as a percentage of true protein of the calibration milks would have an effect on the MIR uncorrected readings, calibration slopes and intercepts, and MIR predicted readings. However, the influence of lipolysis and proteolysis on uncorrected readings was either not significant, or significant but very small. Likewise, the amount of variation accounted for by day of storage at 4°C of a calibration set on the calibration slopes and intercepts was also very small. Most of the variation in uncorrected readings and calibration slopes and intercepts were due to differences between the optical filter and virtual filter analyzers and differences between the pasteurized modified milk and raw producer milk calibration sets, not due to lipolysis or proteolysis. The combined impact of lipolysis and proteolysis on MIR predicted values was <0.01% in most cases.
Key Words: calibration • infrared milk analysis • lipolysis • proteolysis
|
INTRODUCTION
|
|---|
The accuracy of mid-infrared (MIR) milk analysis is affected by instrumental factors (Smith et al., 1993a,b, 1995) and the characteristics of the materials used to calibrate the instruments (Kaylegian et al., 2006a). Variations in instrumental factors are minimized by routine precalibration of MIR milk analyzers to maintain mechanical and electronic performance within tolerances (Lynch et al., 2006). Kaylegian et al. (2006b) reported, based on a calibration performance validation study of MIR milk analyzers, that instruments calibrated with modified calibration samples with all laboratory mean chemistry reference values from a group of laboratories gave better validation accuracy than instruments calibrated with producer milk samples with reference values from a single or small number of laboratories. Currently, most MIR milk analyzers in the United States are calibrated with producer milk, but it is likely that the use of modified milk calibration samples will increase in the future. Therefore, in the current study we evaluated both modified milk and producer milk calibration samples.
Traditionally, samples used for calibration of MIR milk analyzers are preserved, raw individual producer milks (Kaylegian et al., 2006a), which have a shelf life of 15 d when stored at 4°C. An alternative approach is the use of calibration sets made from pasteurized modified milk that are preserved and which have a shelf life of 28 d when stored at 4°C (Kaylegian et al., 2006a). The use of preservatives inhibits milk deterioration due to bacterial growth (Santos et al., 2003) but does not prevent deterioration due to lipolytic and proteolytic enzymes (Robertson et al., 1981; Senyk et al., 1985; Santos et al., 2003; Ma et al., 2003). Calibration may be affected by deterioration of the calibration samples during their shelf life.
Traditional optical filter-based MIR milk analyzers use 4 sample wavelengths to measure fat (fat A and fat B), protein, and lactose, along with 4 corresponding reference wavelengths to account for the effects of water absorption and light scattering. Filter-based MIR instruments use optical filters, whereas more recent MIR instrumentation employs Fourier transform (FT) infrared (FTIR) technology and the use of virtual filters. Most FT MIR milk analyzers can be set up to simulate optical filter instruments by selecting the appropriate sample and reference center wavelengths and bandwidths to create virtual filters and use the fixed-filter analysis approach. The light absorbance at the MIR fat A sample wavelength (5.73 µm) is due to carbonyl ester bonds, and absorbance at the fat B sample wavelength (3.48 µm) is due to the carbon-hydrogen stretch. It would be expected that lipolysis of triglycerides in milk fat would reduce absorbance at the fat A wavelength but not impact fat B. Robertson et al. (1981) compared MIR results from a control milk preserved with potassium dichromate to portions of the same control milk that were homogenized to increase lipolysis and were either unpreserved, or preserved with potassium dichromate or sodium azide. They observed a decrease in the fat A reading of 0.033% fat and an increase in protein of 0.019% protein per µEq of FFA/mL of milk after 336 h of storage at 4°C. Although others (Sjaunja, 1982; Sjaunja and Andersson, 1985; van de Voort et al., 1987) have reported similar results for fat A, the conditions (e.g., prolonged recirculating homogenization and pH adjustment) used in these studies to induce lipolysis may have caused other changes in the milk that could have produced effects on MIR readings at other wavelengths that were not due solely to lipolysis. No reports of the influence of proteolysis in milk on MIR analysis were found in the literature.
Individual milks in the population for testing are susceptible to lipolytic and proteolytic deterioration during typical storage conditions up to 5 d. Preserved calibration milks, either raw or pasteurized, are kept even longer (up to 28 d), and lipolysis and proteolysis could have the additional effect of causing a change in the calibration slope and intercept. The objective of this study was to determine the influence of the lipolysis and proteolysis during storage (4°C) of preserved modified milk and producer milk calibration samples on the fat B, fat A, protein, and lactose uncorrected readings, and calibration slopes and intercepts from a filter-based and an FT MIR milk analyzer.
|
MATERIALS AND METHODS
|
|---|
Experimental Design
Lipolytic and proteolytic deterioration was measured weekly for 3 modified milk calibration sets (1, 8, 15, 21, and 28 d) and for 3 producer milk calibration sets (1, 8, and 15 d) stored at 4°C (Kaylegian et al., 2006a). The total study spanned a 90-d period. The first day of each producer calibration set was timed to start on the first day of the modified milk sets so that each set was approximately the same age at the time of MIR analysis, as described previously in experiment 2 in Kaylegian et al. (2006a). Modified and producer milk sets were used separately to calibrate a fixed-filter and an FT MIR analyzer. The slope and intercept of the calibration linear regression for fat B, fat A, protein, and lactose was determined at 1, 5, 8, 12, 15, 19, 21, 26, and 28 d for the modified milk sets and at 1, 4, 8, 11, and 15 d for the producer milk sets.
Calibration Samples
The 14-sample modified milk sets were manufactured at Cornell University from pasteurized, gravity-separated cream, skim milk UF permeate and retentate, anhydrous lactose, and water (Kaylegian et al., 2006a). The 12-sample raw individual producer milk calibration sets were obtained from a USDA Federal Milk Market Laboratory that normally produces calibration sets (Kaylegian et al., 2006a). Modified milk sets were preserved with potassium dichromate at a concentration of 0.02% (ACS grade, Fisher Scientific, East Lawn, NJ). The producer milk sets were preserved with 0.02% potassium dichromate. The pasteurized modified milk sets had a 28-d shelf life and the raw producer milk sets had a 15-d shelf life at 4°C.
Production of Pilot Milk
To monitor the stability of the instruments over the 90-d experiment period, uncorrected readings (Lynch et al., 2006) for fat B, fat A, protein, and lactose were measured for a homogenized milk pilot sample on each day that the calibration sets were analyzed. Three batches of pilot samples were produced from the same raw whole milk that was used to produce the 3 sets of modified milk calibration samples. Raw whole milk was pasteurized (72°C for 16 s), homogenized (model 75E, Gaulin, Evert, MA) at 60°C with a first stage pressure of 13.8 MPa and a second stage of 3.5 MPa, and then was cooled to 4°C. The pilot milk was preserved with 0.02% potassium dichromate and partitioned into 60 mL vials (Capitol Vial, Fultonville, NY). The pilot samples were frozen in liquid nitrogen on the day they were produced and then stored at –80°C. The pilot milk was held frozen to ensure that no age-related lipolysis and proteolysis occurred. On each day of analysis, 2 pilot milks were immediately thawed in a microwave oven (frozen directly to <10°C), immediately tempered to 40°C in a water bath, and duplicate uncorrected component values were determined on each MIR instrument.
Chemical Analysis
Reference Chemistry for Calibration Sets.
All modified milk and producer milk calibration samples were analyzed using reference chemistry methods by at least 7 laboratories for fat, true protein, and total solids, and by at least 4 laboratories for lactose (Kaylegian et al., 2006a). The all laboratory mean chemistry values were used as the reference chemistry for MIR calibration. Chemical analyses were conducted using the following AOAC (2000) methods: fat by modified Mojonnier ether extraction (method 989.05; 33.2.26), true protein (TP) by Kjeldahl nitrogen analysis (method 991.22; 33.2.13), total solids by oven drying (method 990.20; 33.2.44), anhydrous lactose by enzyme analysis (method 984.15; 33.2.24), and milk SCC was determined using a fluorimetric method (method 978.26; 17.13.01).
Lipolysis.
Lipolytic deterioration during storage at 4°C, measured as the increase in FFA content and expressed as milliequivalents of FFA per kilogram of milk, was determined using a modified copper soap method (Ma et al., 2003). The FFA content of each milk sample in each calibration set was determined once weekly. A linear regression of FFA content of each milk sample as a function of storage time was done to obtain an equation that allowed calculation of the FFA content of each calibration sample at any day during storage. From the FFA content of each sample within a calibration set, a mean FFA value for the full calibration set was calculated for any day during storage.
Proteolysis.
Proteolytic deterioration during storage at 4°C was measured as the decrease in casein as a percentage of TP (CN%TP). The AOAC (2000) Kjeldahl methods used were as follows: TP (method 991.22; 33.2.13), NPN (method 991.21; 33.2.12), and noncasein nitrogen (NCN; method 998.05; 33.2.64). Total nitrogen was calculated as the sum of the TP and NPN (TN = TP + NPN). Casein was calculated by subtracting the NCN from TN (CN = TN – NCN). The CN%TP was calculated as (CN/TP) x 100 and was determined once weekly during the 4°C storage. A linear regression of decrease in CN%TP content of each sample as a function of storage time was done as described above for lipolysis. From the decrease in CN%TP of each sample within a calibration set, a mean decrease in CN%TP value for the full calibration set was calculated for any day during storage.
MIR Milk Analysis
The MIR analyses were performed with an optical filter-based instrument (MilkoScan 605, Foss Electric, Hillerød, Denmark) and a virtual filter instrument (LactoScope FTIR, Delta Instruments, Drachten, the Netherlands) operated to simulate an optical filter instrument. The sample and reference wavelengths used for the MilkoScan 605 were as described by Kaylegian et al. (2006a). The peak wavelength for sample and reference virtual filters for the LactoScope FTIR were as follows: 3.51 and 3.56 µm (2,850 and 2,810 cm–1) for fat B, 5.79 and 5.62 µm (1,735 and 1,789 cm–1) for fat A, 6.60 and 6.77 µm (1,523 and 1,486 cm–1) for protein, and 9.54 and 7.79 µm (1,048 and 1,295 cm–1) for lactose, respectively. The LactoScope FTIR total bandwidths were 0.03, 0.05, 0.07, and 0.20 µm for fat B, fat A, protein and lactose, respectively.
Both instruments were precalibrated monthly to ensure that mechanical and electronic performances were within the specified tolerances (Lynch et al., 2006). The uncorrected readings (as defined by Lynch et al., 2006) and the calibration linear regression slope and intercept data were collected in databases using the IR-QC software that was described by Kaylegian et al. (2006a). Separate calibration databases were used for the modified milk and the producer milk calibration sets for each instrument.
Influence of Lipolysis and Proteolysis on MIR Measurement of Milk Components
To determine the direct effects of lipolysis and proteolysis on the MIR measurement of fat B, fat A, protein, and lactose, the uncorrected readings obtained during calibration of both the MilkoScan 605 and the Lacto-Scope FTIR were used. Mean uncorrected values for fat B, fat A, protein, and lactose were determined over the entire shelf life (i.e., 28 d for modified and 15 d for producer milks) for each sample in the 3 modified and 3 producer milk calibration sets for each instrument. The mean uncorrected reading was subtracted from the uncorrected reading at each day of MIR analysis to yield the residual difference. The uncorrected reading residuals differences for both the MilkoScan 605 and LactoScope FTIR were plotted together for each sample and component as a function of the increase in FFA or decrease in CN%TP. The use of residual plots instead of the uncorrected readings eliminated the shifts observed in uncorrected readings when changing from one calibration set to the next, as reported by Kaylegian et al. (2006a) and allowed the data from all 3 sets and both instruments to be combined on one plot for each set type.
Statistical Analysis
All ANOVA were done using the Proc GLM function of SAS (version 8e, SAS Institute, Cary, NC). Day of storage at 4°C was treated as a continuous variable in all the ANOVA models described below. Distortion of the ANOVA by multicollinearity of the linear and quadratic terms for day of storage was minimized by centering the time of storage using a mathematical transformation (Glantz and Slinker, 2001). The day of storage was transformed as follows: day transformed = day of storage – [(last storage day – first storage day)/2]. This transformation made the data set orthogonal with respect to storage time. The model error term was used as the error term for all tests of significance. The model was considered significant if the F-test had a P 
0.05. If the model was significant, then the least squares means were compared (P 0.05). In all ANOVA tables, the type III sum of squares (SS) for each independent variable was expressed as the percentage of the total variation where percentage variation = (model independent variable type III SS/corrected total SS) x 100.
Lipolysis and Proteolysis of Calibration Sets.
An ANOVA was performed to determine if the mean FFA concentration or mean decrease in CN%TP differed between set types or changed with day of storage. The statistical analysis included the first 15 d of storage of the modified milk and the full 15-d shelf life of the producer milk sets to directly compare the effects of the 2 types of calibration sets during the same period. Class variables were set type (modified milk or producer milk) and replicate (set number 1, 2, or 3).
Influence of Milk SCC on Lipolysis and Proteolysis of Calibration Samples.
An ANOVA was performed to determine the influence of SCC on the increase in FFA content and on the decrease in CN%TP. The values for the increase in FFA content and the decrease in CN%TP at the last day of the calibration for each sample were used (i.e., d 28 for modified milk and d 15 for producer milk sets). Class variables were set type (modified or producer milk) and replicate (set number 1, 2, or 3); SCC was treated as a continuous variable that was transformed for each sample in the set as described above for day of storage.
MIR Instrument Stability.
A regression analysis was performed using the Proc REG function in SAS for the homogenized pilot to determine if there was any change in the uncorrected reading residuals over the 4-wk storage period of each set (i.e., at d 1, 5, 8, 12, 15, 19, 21, 26, and 28). Mean uncorrected values for fat B, fat A, protein, and lactose were determined over the entire shelf life of the 3 pilot batches for each instrument. The mean uncorrected reading was subtracted from the uncorrected reading at each day of MIR analysis to yield the residual difference. The regression analysis was conducted individually for each milk component, and the model used was day of storage = uncorrected reading residual for fat B, fat A, protein or lactose.
Influence of Lipolysis and Proteolysis of Calibration Sets on Uncorrected Readings.
A regression analysis was performed to determine if the uncorrected reading residuals for fat B, fat A, protein, and lactose changed with changes in FFA content or with decrease in CN%TP of the calibration sets. The uncorrected reading residuals were calculated as described above. The regression analysis was conducted individually for each milk component, and the model used was change in FFA content or decrease in CN%TP = uncorrected reading residual for fat B, fat A, protein, or lactose.
Changes in Calibration Slopes and Intercepts.
An ANOVA was performed for each calibration set type for fat B, fat A, protein, and lactose to determine if there were any changes in the linear regression calibration slope and intercept within calibration set type over the entire shelf life (i.e., 28 d for modified and 15 d for producer milk sets). Class variables were instrument (MilkoScan 605 or LactoScope FTIR) and replicate (set number 1, 2, or 3).
|
RESULTS AND DISCUSSION
|
|---|
Lipolysis of Calibration Sets During Storage at 4°C
Lipolysis during storage of preserved milk samples can be caused by lipoprotein lipase in raw milk or heat-resistant microbial or somatic cell-associated lipases in pasteurized milk (Santos et al., 2003). The mean FFA for the modified and producer calibration sets plotted as a function of day of storage at 4°C is shown in Figure 1
. The producer milk calibration sets had a shelf life of 15 d and, therefore, an ANOVA comparison of the first 15 d of the modified milk sets with the full 15 d shelf life of the producer milk sets was done. There was a difference (P 0.01) in mean FFA content between the set types (Table 1), and the producer milk calibration had a higher mean FFA content (least squares mean = 0.253 mEq/kg of milk for the producer milk set and 0.115 mEq/kg of milk for the modified milk set).
View this table:
[in this window]
[in a new window]
|
Table 1. The ANOVA model, degrees of freedom (df), and percentage of total variation explained by each term in the model for the mean FFA content (mEq/kg milk) of 3 modified milk and 3 producer calibration sets over 15 d of storage at 4°C
|
|
The significant day by set type interaction (Table 1
) indicated that the producer milk sets were undergoing lipolysis (i.e., increasing in FFA content) over the 15-d shelf life, but no change in the FFA content of the modified milk sets with time was detected (Figure 1). This was expected because the producer calibration sets were raw and contained active milk lipase, whereas the modified milk sets were pasteurized. Pasteurization (HTST) inactivates most of the native milk lipase (Shipe and Senyk, 1981).
The initial FFA content in the modified milk sets (Table 2) was correlated with the increasing fat content, and as fat content increased incrementally from 0.20 to 5.70% (Kaylegian et al., 2006a), the FFA content increased from about 0.09 to between 0.12 and 0.24 mEq/kg of milk. The source of FFA in the modified milk sets was the gravity-separated cream. The addition of progressively increasing amounts of cream from sample 1 to 14 would be expected to cause the FFA content to increase in direct relationship with fat content. Overall, the FFA content of the modified milk sets was low.
View this table:
[in this window]
[in a new window]
|
Table 2. The FFA content for individual milk samples in 3 modified milk calibration sets at d 1 of storage at 4°C
|
|
In contrast to the modified milk samples (Table 2), the FFA content of the producer milk sets (Table 3) did not show any systematic correlation with fat content within set (data not shown). The producer milk sets had a greater range of the initial FFA content (Table 3) in the 3 sets (0.290, 0.365, and 0.308 mEq/kg of milk, respectively) than the modified milk sets (0.118, 0.119, and 0.161 mEq/kg of milk, respectively; Table 2). There was a large variation in the amount of lipolysis that occurred within producer sets (Table 3). There was also no consistent relationship between the initial FFA content and the increase in FFA over the 15-d shelf life, as illustrated by 2 samples that had a similar initial FFA content (0.426 mEq/kg of milk, Table 3) with an increase of 0.19 mEq/kg of milk in one sample (set 1, sample 7), and 0.107 mEq/kg of milk in the other (set 2, sample 5). Factors known to affect the variation in the FFA content in raw milk include inherent cow-to-cow variation (Tarassuk and Frankel, 1957), agitation and temperature control (Fitz-Gerald, 1974), and mechanical handling (Deeth and Fitz-Gerald, 1978) of raw milk, which causes disruption in the milk fat globule and promotes lipolysis.
View this table:
[in this window]
[in a new window]
|
Table 3. The FFA content for individual milk samples in 3 producer milk calibration sets at d 1 and 15 of storage at 4°C
|
|
Proteolysis of Calibration Sets During Storage at 4°C
Proteolysis can be caused by the native milk protease plasmin (Klei et al., 1997), proteases from somatic cells (Saeman et al., 1988; Verdi and Barbano, 1991b), and microbial proteases (Guinot-Thomas et al., 1995). Proteolysis can occur during 4°C storage of preserved milk samples (Santos et al., 2003). The mean decrease in CN%TP for modified and producer milk calibration sets plotted as a function of days of storage at 4°C is shown in Figure 2. The data for CN%TP content of the individual modified and producer milk calibration sets are shown in Tables 4 and 5, respectively. The sample-to-sample variation in proteolysis within the modified milk sets was less than within the producer milk sets because each modified milk set was made from the same batch of pasteurized milk, whereas the producer milk sets were made from milk from 12 different individual farms and reflected typical farm-to-farm variation in CN%TP (Verdi et al., 1987).
View this table:
[in this window]
[in a new window]
|
Table 4. Casein as a percentage of true protein (CN%TP) for individual milk samples in 3 modified milk calibration sets at 1, 8, 15, 21, and 28 d of storage at 4°C
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. The casein as a percentage of true protein (CN%TP) for individual milk samples in 3 producer milk calibration sets at 1, 8, and 15 d of storage at 4°C
|
|
An ANOVA comparison of the first 15 d of the modified milk sets with the full 15-d shelf life of the producer milk sets was done to determine if proteolysis in the 2 types of calibration milks differed. The modified milk sets (Figure 2a) had a larger (P 0.01, Table 6) mean decrease in CN%TP (LSM = 0.99% for modified milk and 0.81% for producer milk) than the raw producer milk calibration samples. More proteolysis with increasing time of storage in the preserved pasteurized modified milk sets than in preserved raw producer milk sets is consistent with the observations of Santos et al. (2003) and was due to the destruction of plasminogen activator inhibitors by pasteurization (Richardson, 1983).
View this table:
[in this window]
[in a new window]
|
Table 6. The ANOVA model, degrees of freedom (df), and percent variation for each term in the model for the mean decrease in CN%TP of 3 modified milk and 3 producer milk calibration sets over 15 d of storage at 4°C
|
|
Day of storage explained most of the variation in CN%TP (Table 6) in this experiment and reflected the large time-dependent increase (Figure 2) in proteolysis in both types of preserved calibration sets. A significant (P 0.01) interaction of set type with day of storage indicated that the modified milk sets (Figure 2a) underwent proteolysis at a slightly faster rate than the producer milk sets (Figure 2b). At d 15 of storage at 4°C, the mean decrease in CN%TP was 1.96% for the modified milk sets and 1.54% for the producer milk sets. At the end of the 28-d shelf life of the modified milk sets, the mean decrease in CN%TP was 4.1%. Although this was a relatively large amount of proteolysis, there was no visual evidence of coagulation or protein destabilization in the modified milk sets at 28 d. The potassium dichromate preservative prevented microbial growth (Santos et al., 2003); therefore, the observed proteolysis was primarily due to the action of the native milk proteases.
Influence of Milk SCC on Lipolysis and Proteolysis of Calibration Samples
The 3 modified milk sets were made with raw whole milks that had low SCC (256,000, 260,000, and 270,000 cells/mL for sets 1, 2, and 3, respectively). The SCC of the modified milk sets (Table 7) had a systematic increase from sample 1 to 14 from as low as 19,000 to as high as 804,000 cells/mL, which was directly correlated with the increase in fat content (Kaylegian et al., 2006a). The somatic cells partitioned with the cream during gravity separation of pasteurized milk during the manufacture of the modified milk sets (Kaylegian et al., 2006a). Therefore, modified milk samples with higher fat contents (i.e., increased cream content) had higher SCC than did low-fat samples. The producer milk sets had a SCC within sets that ranged from as low as 85,000 to as high as 800,000 cells/mL (Table 7) due to farm-to-farm variation, but SCC was not correlated with fat content (data not shown).
Lipolysis.
In other studies, raw milk with high SCC was shown to have a higher FFA content (Fitz-Gerald et al., 1981; Bachman et al., 1988; Ma et al., 2000) and lipase activity (Azzara and Dimick, 1985a,b) compared with milk with low SCC. Pasteurized preserved milk with high SCC had higher FFA content after 29 d of storage at 0.5°C and 6°C than low SCC milk (Santos et al., 2003).
An ANOVA was done to determine if there was any impact of milk SCC on lipolysis during storage in modified and producer milk sets. As expected, there was an effect of set type (P 0.01, Table 8) because the producer milk sets had a higher initial FFA content and increased in FFA content during storage, whereas the FFA content of the modified milk sets did not (Figure 1). No impact (P > 0.05) was detected of milk SCC or the interaction of SCC with set type on the increase in FFA content during storage at 4°C for 15 d for the producer milk sets or for 28 d for the modified milk sets. Although the SCC of the modified milks had a wide range, the somatic cells originated from low SCC raw milk. It is likely that the low SCC raw milk used to make the modified milk sets contained a lower proportion of neutrophils and more macrophages and lymphocytes (Azzara and Dimick, 1985b; Sarikaya et al., 2004) than typically found in high SCC milk. This, together with the fact that the modified milk samples were pasteurized, may partially explain why the FFA content of the modified milk samples did not increase with time (Figure 1), even in the modified milk samples that had high SCC (e.g., samples 13 and 14, Table 7).
View this table:
[in this window]
[in a new window]
|
Table 8. The ANOVA model, degrees of freedom (df), and percent variation for each term in the model for the increase in FFA content (meq/kg milk) and decrease in casein as a percentage of true protein (CN%TP) at the last day of storage (4°C) as a function of SCC in the 3 modified milk (at d 28) and 3 producer milk (at d 15) calibration sets
|
|
Proteolysis.
There are many reports in the literature of a high correlation between milk with high SCC and increased proteolysis (Andrews, 1983; Senyk et al., 1985; Verdi et al., 1987; Verdi and Barbano, 1988, 1991a, Verdi and Barbano, b). An ANOVA was done to determine if there was any impact of milk SCC on proteolysis during storage at 4°C within modified and producer milk calibration sets. Most of the variation in decrease in CN%TP was due to set type (Table 8), with the modified milks having overall higher levels of decrease in CN%TP (Figure 3a) than producer milks (Figure 3b).
View larger version (17K):
[in this window]
[in a new window]
|
Figure 3. Decrease in casein as a percentage of true protein (CN%TP) at the last day of shelf life as a function of SCC in (a) modified milk at d 28; and (b) producer milk at d 15.
|
|
There was a significant (P 0.01; Table 8) influence of both SCC and the interaction of SCC by set type on the decrease in CN%TP. The significant interaction of SCC by set type can be seen by comparing Figures 3a and 3b. No influence (P > 0.05) of SCC on the decrease in CN%TP in the modified milk samples (Figure 3a) was detected. The somatic cells in the modified milk samples originated from low SCC raw milk and this maybe why CN%TP did not change with SCC (Figure 3a). The pattern of proteolytic breakdown of casein has been previously reported to be different in low SCC and high SCC milks (Saeman et al., 1988; Verdi and Barbano, 1991b). A significant (P 0.01) influence of increasing SCC on the decrease in CN%TP (i.e., more proteolysis) for the producer milks (Figure 3b) was observed. The higher SCC producer milks were from groups of cows with higher levels of mastitis than those producing the bulk milk used to make the modified milk sets. The producer milks with high SCC would be expected to have more proteolysis (Verdi et al., 1987; Saeman et al., 1988) than the modified milks with high SCC.
Stability of Pilot Sample Uncorrected Readings
Unlike the modified and producer milks, the pilot milks were kept frozen (–80°C) so that there would be no proteolysis or lipolysis during storage. These pilot milks served as a control to determine if each instrument would give the same reading within each of the three 4-wk periods during the study. No significant (P > 0.05) change in the fat B, fat A, or protein readings on the pilot milks was detected during any of the 4-wk periods for the MilkoScan 605 or the LactoScope FTIR (data not shown). This indicated that the fat B, fat A, and protein readings of milk that did not undergo proteolysis or lipolysis did not change. For lactose, no change in the reading on the pilot milks was detected on the LactoScope FTIR, but there was a significant (P 0.01) change over each 4-wk period on the MilkoScan 605. The lactose reading on the MilkoScan 605 gradually increased (cause unknown) during each 4-wk period, but the magnitude of the increase in lactose was small (about 0.02 to 0.03% lactose). Overall, both instruments were very stable during the study period.
Influence of Lipolysis and Proteolysis of Calibration Samples on Uncorrected Readings
Lipolysis.
Lipolysis in producer milk sets increased with time of storage (Figure 1, Table 3); this was not observed in the modified sets (Figure 1). Therefore, data from the 2 samples that had the largest increase in FFA from each of the 3 producer milk sets (samples 1 and 9 from set 1, samples 5 and 9 from set 2, and samples 2 and 12 from set 3) were used to determine the effects of lipolysis on MIR uncorrected readings (Figure 4). No significant (P > 0.05) change in fat B, fat A, and lactose readings (Figures 4a, b, and d, respectively) of the raw, preserved producer milks for both the optical and virtual filter instruments over 15-d storage at 4°C was detected with increasing FFA content. There was no influence of lipolysis for either instrument, so the residuals for both instruments were plotted together in Figure 4.
View larger version (11K):
[in this window]
[in a new window]
|
Figure 4. Uncorrected reading residuals as a function of the increase in FFA content during 15 d storage at 4°C for the 2 producer milk calibration samples with the highest lipolysis from each replicate (6 samples total): (a) fat B, (b) fat A, (c) protein, and (d) lactose. Slope of residuals was different (P 0.05) from zero for protein.
|
|
There was a significant increase (P 0.05) in the protein uncorrected reading (Figure 4c) with an increase in FFA content of 0.20 mEq/kg of milk; however, the magnitude of the increase in the uncorrected protein reading was small (about 0.01%). The increase in protein reading was due to the absorbance of the carboxylate anions of dissociated FFA (Silverstein and Bassler, 1967) in the wavelength range of 6.06 to 6.45 µm (1,650 to 1,550 cm–1) near the region of the protein sample filter. Our data for higher MIR protein readings as FFA content increased are consistent with other reports (Robertson et al., 1981; Kerkhof Mogot et al., 1982; van de Voort et al., 1987).
Proteolysis.
Proteolysis in the modified milk sets increased with time of storage (Figure 2a) more than for the producer milk sets (Figure 2b). Therefore, the decrease in CN%TP for the 2 milks that had the largest decrease in CN%TP (3.63 to 5.52% at d 28) from each modified milk set (samples 2 and 8 from set 1, samples 1 and 12 from set 2, and samples 7 and 8 from set 3) were used to determine the effects of proteolysis on MIR uncorrected readings (as defined by Lynch et al., 2006). There was a significant (P 0.05) increase in uncorrected fat B, fat A, and protein readings with decreasing CN%TP (Figures 5a, b, and c, respectively), but the increase in MIR uncorrected reading was small (0.01 to 0.02% fat or protein). The impact of proteolysis was the same for both instruments, so the residuals for both instruments are plotted together in Figure 5. No effect (P > 0.05) of proteolysis on the lactose uncorrected reading was detected. It was surprising that this high level of proteolysis had so little impact on MIR uncorrected readings. From a practical point of view, it is likely that higher levels of proteolysis than observed in this study would cause disruption of the physical stability of the milk, which would be obvious to an analyst by visual inspection.
View larger version (15K):
[in this window]
[in a new window]
|
Figure 5. Uncorrected reading residuals as a function of the decrease in casein as a percentage of true protein (CN%TP) during 28 d of storage at 4°C for the 2 modified milk calibration samples with highest proteolysis from each replicate (6 samples total): (a) fat B, (b) fat A, (c) protein, and (d) lactose. Slope of residuals different (P 0.05) from zero for fat B, fat A, and protein.
|
|
Change in Calibration During Calibration Set Storage at 4 °C
The uncorrected MIR readings are the MIR signals after linearity and primary slope are adjusted and prior to the application of intercorrection factors and a secondary slope and intercept. Control of the uncorrected signal by controlling linearity and primary slope, as described by Lynch et al. (2006), stabilizes the intercorrection factors for an instrument. Intercorrection factors are determined during precalibration (Lynch et al., 2006) and are not expected to vary unless something fundamental changes in the instrument (e.g., the filter is replaced). Applying intercorrection factors to the uncorrected readings produces an intercorrected reading. Calibration of MIR instruments is the process of using linear regression to establish a slope and intercept to convert the intercorrected reading to a corrected reading (predicted chemistry) in the fixed-filter approach. Calibration is accomplished using a set of calibration milks with reference values assigned by chemical analysis. There are 2 channels for measurement of fat, fat B and fat A. Most laboratories in the UDSA Federal Milk Markets use 100% fat B for fat testing of raw milk and do not use fat A.
Calibration slope and intercept are influenced by instrumental factors as well as the characteristics of the milks in the calibration sets. Changes in slope and intercept are observed when switching from one calibration set to another (Kaylegian et al., 2006a), especially when the calibration milks are producer-based. If the same calibration set is used over the shelf life of the calibration samples, the slope and intercept should technically stay the same. However, this would only hold true if the uncorrected MIR signals were unaffected by age-related changes in the calibration milks.
The effects of age-related lipolysis and proteolysis of the modified and producer calibration milks on calibration slopes and intercepts were evaluated at each component channel. The ANOVA model included instrument type (optical and virtual filter-based), day of storage (28 and 15 d for the modified and producer milks, respectively), and replicate (3 replicates per calibration type). The influence of instrument type and day of storage as a percentage of the total variation is shown for calibration slopes in Table 9 and calibration intercepts in Table 10. Most of the variation in slope (Table 9) and intercept (Table 10) was explained by the instrument type (P 0.05) for both the modified and producer milk calibration sets (P 0.05). This was expected because of differences in the gain and intercorrection factors (Lynch et al., 2006) between the MilkoScan 605 and LactoScope FTIR caused by differences in the filters (i.e., center wavelength and bandwidth). The effects of replicate and the interaction between replicate and instrument type were significant (P < 0.05) from one calibration set to the next (Tables 9 and 10). Shifts in the calibration slope and intercept values when changing from one set of calibration samples to the next, particularly for producer calibration sets, are common (Kaylegian et al., 2006a), and this is why the replicate effect and replicate by instrument effect were significant in several cases (Tables 9 and 10). For the modified milk sets, there was a small but significant influence of day of storage or a day x instrument interaction on and fat B and fat A slopes (Table 9) and fat B, fat A, and protein intercepts (Table 10).
View this table:
[in this window]
[in a new window]
|
Table 9. The ANOVA model, degrees of freedom (df), and percentage variation explained by each term in the model for fat B, fat A, protein, and lactose slopes over the 28-d shelf life of modified milk calibration sets and over the 15-d shelf life of producer milk calibration sets
|
|
View this table:
[in this window]
[in a new window]
|
Table 10. The ANOVA model, degrees of freedom (df), and percent variation explained by each term in the model for fat B, fat A, protein, and lactose intercepts over the 28 d shelf life of modified milk calibration sets and over the 15 d shelf life of producer milk calibration sets
|
|
The change in slopes and intercepts for fat B, fat A, protein, and lactose from d 1 to 28 for the modified milk sets are shown in Table 11 and from d 1 to 15 for the producer milk sets are shown in Table 12. When the effect of day of storage and interactions with day of storage were significant, they explained <5.6% of the variation in slope (Table 9) for the modified milk sets and <1.1% for the producer milk sets, and <1.45 and <5.9% of the variation in intercepts (Table 10) for the modified milk and producer milk sets, respectively. The impact of these combined slope and intercept changes on the MIR predicted fat, protein, and lactose for an average milk sample (using intercorrected values of 3.60% fat, 3.05% protein, and 4.55% anhydrous lactose) are shown as change in MIR predicted values in Table 11 for modified milks and Table 12 for producer milks. There was a significant day or day x instrument interaction for fat B, fat A, and protein for the modified sets. The change in predicted value was small, <0.01% fat over 28 d for fat B (Table 11). For fat A, more variation in slope with day of storage was explained by the day x instrument interaction term than by the day term (Table 9). The effect of this interaction can be seen in Table 11 where the LactoScope FTIR had a larger change in predicted value (–0.022%) than the MilkoScan 605 (–0.002%). However, fat A alone is generally not used for measuring fat for payment testing, but is used in combination with fat B (Biggs and McKenna, 1989). The effect of day of storage on protein produced a change in MIR-predicted value of just greater than 0.01% protein on both instruments for the modified milks (Table 11). The change in MIR predicted values for the producer milk set over 15 d of storage are shown in Table 12. There was a significant day or day x instrument interaction for fat B and fat A slopes and intercepts for producer milks (Table 9). The predicted change in fat B and fat A MIR-predicted value over 15 d of storage for producer milks was <0.01% for both instruments (Table 12). Therefore, considering all sources of variation, the impact of lipolysis and proteolysis on MIR predicted milk component values was small.
View this table:
[in this window]
[in a new window]
|
Table 11. Change in calibration slope and intercept and change in mid-infrared (MIR) predicted value for fat B, fat A, protein, and lactose when using the MilkoScan 605 and LactoScope FTIR from d 1 to 28 for modified milk sets
|
|
View this table:
[in this window]
[in a new window]
|
Table 12. Change in calibration slope and intercept and change in MIR predicted value for fat B, fat A, protein, and lactose when using the MilkoScan 605 and LactoScope FTIR from d 1 to 15 for producer milk sets
|
|
|
CONCLUSIONS
|
|---|
The focus of this study was to determine if lipolysis or proteolysis during the shelf life of calibration sets influenced the MIR readings or calibration slopes and intercepts. It was expected that the increases in FFA and decreases in CN%TP of the calibration milks observed in this study could have an effect on the MIR uncorrected readings or calibration slopes and intercepts. However, the influence of lipolysis and proteolysis observed in the uncorrected readings was either not significant or significant (P 0.05) but very small. The amount of variation accounted for by day of storage at 4°C of a calibration set on calibration slopes and intercepts was also very small (<6%). Most of the variations in uncorrected readings, calibration slopes and intercepts, and MIR-predicted component values observed in this study were due to differences between the MilkoScan 605 and LactoScope FTIR milk analyzers and differences between the characteristics (i.e., component concentration range and distribution of the samples within the range) of the pasteurized modified milk sets and raw producer milk sets, not to proteolysis or lipolysis.
|
ACKNOWLEDGEMENTS
|
|---|
The authors would like to thank all the USDA Federal Milk Market laboratories and affiliated laboratories for their collaboration and sample analysis in this work. The technical assistance in sample preparation by Maureen Chapman, Bob Kaltaler, and Mark Schweisthal was important for the success of this project. The authors thank the Test Procedures Committee of the USDA, Dairy Programs, Federal Milk Markets for their financial support of this research.
|
FOOTNOTES
|
|---|
1 Use of names, names of ingredients, and identification of specific models of equipment is for scientific clarity and does not constitute any of endorsement of product by authors, Cornell University, or the Northeast Dairy Foods Research Center. 
Received for publication March 25, 2006.
Accepted for publication July 24, 2006.
|
REFERENCES
|
|---|
Andrews, A. T. 1983. Breakdown of caseins by proteinases in bovine milks with high somatic cell counts arising from mastitis or infusion with bacterial endotoxin. J. Dairy Res. 50:57–66.[Medline]
AOAC. 2000. Official Methods for Analysis. Assoc. Offic. Anal. Chem. Int., Gaithersburg, Maryland.
Azzara, C. D., and P. S. Dimick. 1985a. Lipolytic enzyme activity of macrophages in bovine mammary gland secretions. J. Dairy Sci. 68:1804–1812.[Abstract/Free Full Text]
Azzara, C. D., and P. S. Dimick. 1985b. Lipoprotein lipase activity of milk from cows with prolonged subclinical mastitis. J. Dairy Sci. 68:3171–3175.[Abstract/Free Full Text]
Bachman, K. C., M. J. Hayen, D. Morse, and C. J. Wilcox. 1988. Effect of pregnancy, milk yield, and somatic cell count on bovine milk fat hydrolysis. J. Dairy Sci. 71:925–931.[Abstract/Free Full Text]
Biggs, D. A., and D. McKenna. 1989. Alternative methods for infrared analysis of fat in milk: Interlaboratory study. J. AOAC 72:724–734.
Deeth, H. C., and C. H. Fitz-Gerald. 1978. Effects of mechanical agitation of raw milk on the milk-fat globule in relation to the level of induced lipolysis. J. Dairy Res. 45:373–380.
Fitz-Gerald, C. H. 1974. Milk lipase activation by agitation—influence of temperature. Aust. J. Dairy Technol. 29:28–32.
Fitz-Gerald, C. H., H. C. Deeth, and B. J. Kitchen. 1981. The relationship between the levels of free fatty acids, lipoprotein lipase, carboxylesterase, N-acetyl-ß-D-glucosaminidase, somatic cell count and other mastitis indices in bovine milk. J. Dairy Res. 48:253–265.[Medline]
Glantz, S. A., and B. K. Slinker. 2001. Multicolinearity and what to do about it. Pages 185–187 in Primer of Applied Regression and Analysis of Variation. 2nd ed. McGraw-Hill, Inc., New York, NY.
Guinot-Thomas, P., M. Al Ammoury, Y. Le Roux, and F. Laurent. 1995. Study of proteolysis during storage of raw milk at 4°C: Effect of plasmin and microbial proteases. Int. Dairy J. 5:685–697.
Kaylegian, K. E., G. E. Houghton, J. M. Lynch, J. R. Fleming, and D. M. Barbano. 2006a. Calibration of infrared milk analyzers: Modified milk versus producer milk. J. Dairy Sci. 89:2817–2832.[Abstract/Free Full Text]
Kaylegian, K. E., J. M. Lynch, G. E. Houghton, J. R. Fleming, and D. M. Barbano. 2006b. Modified versus producer milk calibration: Mid-infrared analyzer performance validation. J. Dairy Sci. 89:2833–2845.[Abstract/Free Full Text]
Kerkhof Mogot, M. F., J. Koops, R. Neeter, K. J. Slangen, H. van Hemert, O. Kooyman, and W. Wooldrik. 1982. Routine testing of farm tank milk with the Milko-Scan 203. I. Calibration procedure and small-scale experiments. Neth. Milk Dairy J. 36:115–133.
Klei, L. R., J. M. Lynch, D. M. Barbano, P. A. Oltenacu, A. J. Lednor, and D. K. Bandler. 1997. Influence of milking three times a day on milk quality. J. Dairy Sci. 80:427–436.[Abstract]
Lynch, J. M., D. M. Barbano, M. Schweisthal, and J. R. Fleming. 2006. Precalibration evaluation procedures for mid-infrared milk analyzers. J. Dairy Sci. 89:2761–2774.[Abstract/Free Full Text]
Ma, Y., D. M. Barbano, D. M. Galton, M. A. Rudan, and K. J. Boor. 2000. Effects of somatic cell count on quality and shelf-life of pasteurized fluid milk. J. Dairy Sci. 83:264–274.[Abstract]
Ma, Y., D. M. Barbano, and M. Santos. 2003. Effect of CO2 addition to raw milk on proteolysis and lipolysis at 4°C. J. Dairy Sci. 86:1616–1631.[Abstract/Free Full Text]
Richardson, B. C. 1983. The proteinases of bovine milk and the effect of pasteurization on their activity. N.Z. J. Dairy Sci. Technol. 18:233–245.
Robertson, N. H., A. Dixon, J. H. Nowers, and D. P. S. Brink. 1981. The influence of lipolysis, pH and homogenization on infra-red readings for fat, protein and lactose. S. Afr. J. Dairy Technol. 13:3–7.
Saeman, A. I., R. J. Verdi, D. M. Galton, and D. M. Barbano. 1988. Effect of mastitis on proteolytic activity in bovine milk. J. Dairy Sci. 71:505–512.[Abstract/Free Full Text]
Santos, M. V., Y. Ma, and D. M. Barbano. 2003. Effect of somatic cell count on proteolysis and lipolysis in pasteurized fluid milk during shelf-life storage. J. Dairy Sci. 86:2491–2503.[Abstract/Free Full Text]
Sarikaya, H., C. Prgomet, M. W. Pfaffl, and R. M. Bruckmaier. 2004. Differentiation of leukocytes in bovine milk. Milchwissenschaft 59:586–589.
Senyk, G. F., D. M. Barbano, and W. F. Shipe. 1985. Proteolysis in milk associated with increasing somatic cell counts. J. Dairy Sci. 68:2189–2194.[Abstract/Free Full Text]
Shipe, W. F., and G. F. Senyk. 1981. Effects of processing conditions on lipolysis in milk. J. Dairy Sci. 64:2146–2149.[Abstract/Free Full Text]
Silverstein, R. M., and G. C. Bassler. 1967. Carboxylate anion. Page 91 in Spectrometric Identification of Organic Compounds. John Wiley and Sons, Inc., New York, NY.
Sjaunja, L. 1982. Studies on milk analyses of individual cow samples. III. Effect of different treatments on infrared analyses. Acta Agric. Scand. 34:273–285.
Sjaunja, L. O., and I. Andersson. 1985. Laboratory experiments with a new infrared (IR) milk analyzer, the Milko-Scan 605. Acta Agric. Scand. 35:345–352.
Smith, E. B., D. M. Barbano, J. M. Lynch, and J. R. Fleming. 1993a. Performance of homogenizers in infrared milk analyzers: A survey. J AOAC Int. 76:1033–1041.
Smith, E. B., D. M. Barbano, J. M. Lynch, and J. R. Fleming. 1993b. A quantitative linearity evaluation method for infrared milk analyzers. J AOAC Int. 76:1300–1308.
Smith, E. B., D. M. Barbano, J. M. Lynch, and J. R. Fleming. 1995. Infrared analysis of milk: Effect of homogenizer and optical filter selection on apparent homogenization efficiency and repeatability. J AOAC Int. 78:1225–1233.
Tarassuk, N. P., and E. N. Frankel. 1957. The specificity of milk lipase. IV. Partition of the lipase system in milk. J. Dairy Sci. 40:418–430.[Abstract/Free Full Text]
van de Voort, F. R., S. Kermasha, J. P. Smith, B. L. Mills, and K. F. Ng-Kwai-Hang. 1987. A study of the stability of record of performance milk samples for infrared milk analysis. J. Dairy Sci. 70:1515–1523.[Abstract/Free Full Text]
Verdi, R. J., and D. M. Barbano. 1988. Preliminary investigation of the properties of somatic cell proteases. J. Dairy Sci. 71:534–538.[Abstract/Free Full Text]
Verdi, R. J., and D. M. Barbano. 1991a. Effect of coagulants, somatic cell enzymes, and extracellular bacterial enzymes on plasminogen activation. J. Dairy Sci. 74:772–782.[Abstract]
Verdi, R. J., and D. M. Barbano. 1991b. Properties of proteases from milk somatic cells and blood leukocytes. J. Dairy Sci. 74:2077–2081.[Abstract]
Verdi, R. J., D. M. Barbano, M. E. Dellavalle, and G. F. Senyk. 1987. Variability in true protein, casein, nonprotein nitrogen, and proteolysis in high and low somatic cell milks. J. Dairy Sci. 70:230–242.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
K. E. Kaylegian, J. M. Lynch, J. R. Fleming, and D. M. Barbano
Influence of fatty acid chain length and unsaturation on mid-infrared milk analysis
J Dairy Sci,
June 1, 2009;
92(6):
2485 - 2501.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
K. E. Kaylegian, D. A. Dwyer, J. M. Lynch, D. E. Bauman, J. R. Fleming, and D. M. Barbano
Impact of fatty acid composition on the accuracy of mid-infrared fat analysis of farm milks
J Dairy Sci,
June 1, 2009;
92(6):
2502 - 2513.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
) set 1, (
) set 2, and (•) set 3; and producer milk calibration sets for d 15: (
) set 1, (
) set 2, and (
) set 3.