Multivariate evaluation of 1-dimensional sarcoplasmic protein profile patterns of turkey breast muscle during early post-hatch development

MOLECULAR, CELLULAR, AND DEVELOPMENTAL BIOLOGY Multivariate evaluation of 1-dimensional sarcoplasmic protein profile patterns of turkey breast muscle during early post-hatch development I. Zapata,* J. M. Reddish,* M. S. Lilburn,† and M. Wick*1 *Department of Animal Sciences, The Ohio State University, Columbus 43210; and †Ohio Agricultural Research and Development Center, The Ohio State University, Wooster 44691 restrict fed. Proteomic PCA plots showed a visual developmental pattern from 7 until 17 d. Multivariate ANOVA highlighted the effect of time point and feeding regimen among profile patterns. The use of different genotypes and feeding regimens influenced variability, which was measured by mean Euclidean distances and ellipses of the PCA plots. These treatment effects, however, did not mask the developmental patterns. After 17 d, the proteomic patterns converged, suggesting that a level of biological stability was achieved regardless of the genotype or treatment. The developmental pattern obtained by the PCA methodology can aid in the planning of more efficient experimental designs so the developmental stage of individuals can be more accurately assessed.

Key words: turkey, development, feed restriction, proteomic, principal component analysis 2011 Poultry Science 90:2828–2836 doi:10.3382/ps.2011-01376

INTRODUCTION Sarcomeric proteins are the primary components of metabolic pathways; thus, proteomic profiles may provide some insight into the metabolic status of an organism. However, proteome analysis and proteomic data interpretation is complicated due to the large number of proteins found within a cell or group of cells. It is often necessary to use discriminative statistical methods to identify which proteins or protein clusters are associated with a specific trait or metabolic phenomenon. Protein pattern recognition is also complicated due to the complex interactions that invariably manipulate biological processes (Barrett et al., 2005). The incorporation of statistical tools into proteome analysis has allowed for the development of structured methods, which facilitate proteomic data interpretation. It has been reported that 1-dimensional electrophoresis coupled with principal component analysis (PCA) is ©2011 Poultry Science Association Inc. Received January 20, 2011. Accepted August 18, 2011. 1 Corresponding author: [email protected]

an effective method for proteomic pattern recognition (Supek et al., 2008). For example, PCA has been used to analyze the associations of numerous variables in a multivariate approach to describe the sensory analysis of deboned chicken breast (Liu et al., 2004), and the evaluation of microbial community structures within poultry litter (Lovanh et al., 2007; Cressman et al., 2010). Breast meat production is the primary economic drive for the turkey industry; the tandem effects of genetic selection and nutrition on BW gain and breast muscle development have been extensively studied (Bayyari et al., 1997; Christensen et al., 1999). Turkey breast muscle was used in the current experiment to illustrate the functionality of the developed proteomic methodology. A study (Havenstein et al., 2007) was specifically designed to address the interactive effects between diet and genotype on the growth and development of a slow-growing random-bred turkey line representative of commercial turkeys available in the mid-1960s (RBC2) and a subline of the RBC2 that was selected at 16 wk of age for BW alone (F line; Nestor et al., 2008). Selection of the F line was initiated in 1969, and there have been numerous studies conducted over

2828

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

ABSTRACT Proteins are the main participants in metabolic pathways. However, the analysis of protein abundance patterns associated with those pathways is complicated by the large number of proteins involved. In this study, the objective was to present the application of principal component analysis (PCA) to permit the visualization of developmental proteomic patterns of sarcoplasmic proteins found in breast muscle. Different turkey genotypes and nutritional regimens were used to potentially increase the variability within the sarcoplasmic protein profile. Sarcoplasmic protein fractions from turkey breast muscle samples were collected at 6 ages between 7 to 24 d. Breast muscle samples were collected from 2 distinctly different turkey lines. The poults within each line were either ad libitum or

TURKEY SARCOPLASMIC PROFILE PATTERN

MATERIALS AND METHODS Line, Diet, and Experimental Design Turkey poults from the RCB2 and F lines were used in this experiment. The poults from each line were reared in heated Petersime battery brooders (Gettysburg, OH). All of the birds were fed at the Ohio State University (Columbus) a turkey starter diet, which met or exceeded the NRC nutrient recommendations for turkeys (NRC, 1994). Poults were allowed either ad libitum access to feed or given 30 min of access to feed per day beginning at 4 d of age (restrict fed). At the beginning of the experiment, 60 RBC2 and 60 Fline poults were randomly assigned to 20 pens (10 pens

for RBC2 and 10 pens for F line). Of the 10 pens assigned to each turkey line, 5 pens housed poults with ad libitum access to feed, whereas the other 5 pens housed poults that were restrict fed. To accommodate for replication and to avoid having each pen represent a single experimental unit, 1 bird per pen was randomly selected from each of the 20 pens at each sampling age for the respective line and treatment combination. During the course of the experiment, some animals died, so the samples collected at later ages represented a reduced number of poults. At the end of the experiment, samples from 56 RBC2 poults and 56 F-line poults were used for the analysis. At 7, 11, 14, 17, 21, or 24 d of age, poults from each genetic line and feed treatment were individually weighed and killed with CO2 (FASS, 1998). Muscle samples were removed from the pectoralis major breast muscle, flash frozen in liquid nitrogen, and subsequently transferred to a −80°C freezer until further analysis. In summary, the experimental design was a 2 (genotype) × 2 (diet) × 6 (age) factorial arrangement of treatments with 4 to 5 biological replicates within each treatment group. All of the animals were handled in compliance with Institutional Animal Care and Use Committee (IACUC) policies and guidelines at The Ohio State University (Columbus).

1-Dimensional Proteomic Profiling The sarcoplasmic fraction of muscle tissue was prepared by homogenizing 250 mg of frozen tissue in 2 mL of rigor buffer [10 mM trismaleate, 60 mM KCl, 5 mM MgCl2, 1 mM EGTA, and 0.4 mM Pefabloc SC Plus, (Boehringer Mannheim Corp., Indianapolis, IN), pH 6.8; Ishiwata and Funatsu, (1985)] on ice. The homogenate was centrifuged at 10,000 × g for 5 min at 4°C. The supernatant was saved and recentrifuged under the same conditions. The supernatant obtained after the second centrifugation step was designated as the sarcoplasmic fraction. Five hundred microliters of the sarcoplasmic fraction was combined with 1 mL of sample buffer (8 M urea/2 M thiourea, 75 mM dithiothreitol, 50 mM Tris, 3% SDS, 0.004% bromophenol blue, pH 6.8) and kept in ice for 30 min. For the 1-dimensional electrophoretic separation, samples were centrifuged again at 10,000 × g for 10 min at 4°C before being loaded onto a discontinuous polyacrylamide slab gel (1 mm × 12 cm × 14 cm) consisting of a 10% resolving gel [30:0.8, acrylamide/N,Nbis(methylene acrylamide)] and a 3% stacking gel. Electrophoretic separation was carried out at a constant voltage of 10 V/cm. After electrophoresis, the gels were stained with SYPRO Ruby protein gel stain (Bio-Rad Laboratories Inc., Hercules, CA) for 24 h, and protected from light. The gels were subsequently destained in 10% methanol and 7% acetic acid for 30 min, and rinsed with deionized water before imaging according to the manufacturer’s recommendation. Imaging was performed as described previously (Zapata et al., 2009)

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

the last 30 yr comparing the RBC2 and F line turkeys for various aspects of growth, reproductive fitness, and disease resistance (Nestor et al., 1999a; Nestor et al., 1999b). The F-line turkeys weigh approximately twice as much as the RBC2 at selection age (16 wk), but there are no correlated increases in the proportional weight of the breast muscles. There are, however, line differences in the expression of key proteins associated with breast muscle growth and development (Updike et al., 2005). One critical protein that is relevant to muscle development is myosin. Myosin has been reported to go through developmental isoform transitions during early growth in chickens (Tidyman et al., 1997) and turkeys (Maruyama and Kanemaki, 1991), and the temporal expression of these isoforms can be influenced by both genetics and nutritional regimen (Graham, 2004; K. Huffman, M. Wick, and N. S. St. Pierre, from the Department of Animal Sciences, The Ohio State University, Columbus, OH, and M. Lilburn from Ohio Agricultural Research and Development Center, The Ohio State University, Wooster, OH, unpublished data). Based on the previous reports of distinct myosin-isoform transition patterns, it was hypothesized that the expression of breast muscle sarcomeric proteins would go through similar temporal transitions. These transitions would be reflected in differences in protein abundance that could be observed by 1-dimensional electrophoresis. In order to use proteomic data analyses for the purpose of improving animal production, it is necessary to develop tools for visualizing the proteomic status of animals from different genotypes and nutritional environments. With this rationale in mind, a methodology for the visual analysis of proteomic data during turkey muscle growth in poults was developed. Thus, the objective of this study was to present a statistical application that would permit the visualization of proteomic developmental patterns of breast muscle sarcoplasmic proteins during the first 24 d of age. Different genotypes and nutritional regimens were included in the study to potentially increase the developmental variability in the sarcoplasmic protein profile.

2829

2830

Zapata et al.

on a Typhoon 9410 laser scanner (GE Healthcare, Chalfont St. Giles, UK). Digital images were analyzed using the Total Lab TL120 software (Nonlinear Dynamics Inc., Newcastle upon Tyne, UK). Bands were matched across samples and gels, and assigned a band index number starting from the top of the gel. Band intensity was measured and expressed as a band percentage of the lane total to normalize for uneven sample loading.

Statistical Analysis

BWijkl = Linei | Treatmentj | Dayk + εijkl, where BWijkl is the dependent variable measured on a poult of Linei (i = RBC2 or F) from Treatmentj [j = full (ad libitum) fed or restrict fed] on Dayk (time point k = 7, 11, 14, 17, 21, and 24 d) from bird l. Linei is the effect of the turkey line; Treatmentj is the effect of the treatment; Dayk is the effect of the day; and εijkl is the random error inherent to each measurement, which is assumed to be independent of other observations and normally distributed with mean zero and SD of σ2. The vertical lines in the equation signify all of the possible interactions. 1-Dimensional Proteomic Profile Analysis. Data were exported from the image analysis software into SAS software (version 9.2, SAS Institute Inc.). The PCA method was performed using the correlation matrix of the band percentages of the lane total as the variables to be analyzed by the PRINCOMP procedure. From the data obtained from the PCA, a component loading plot of the variables (bands) and score plots of the observations (individual samples) were generated by the GPLOT procedure. A further decomposition of the score plots and the addition of 95% confidence ellipses were obtained through the SGPLOT procedure. Principal Component Score Pattern Analysis. To analyze the variation among the principal component score clusters obtained from the analysis of the proteomic data, a second multivariate approach was used by multivariate ANOVA (MANOVA). Component scores from the entire data set were used as a secondary data set. Scores were set as dependent variables, whereas the coded variables of turkey line, dietary treatment, age, and all of their interactions were set as independent variables. The MANOVA analyses were performed in SAS using the GLM procedure. Two separate analy-

RESULTS Weight Gain Mean BW of the RBC2 and F-line poults from each feeding treatment are presented in Figure 1. The F-line poults fed ad libitum weighed significantly more than the RBC2 poults at all ages (P < 0.05). The line differences observed are consistent with previous reports for these 2 lines (Nestor, 1984; Nestor et al., 2000, 2008). Body weight in both lines was significantly reduced by feed restriction (P < 0.05), but within the feed restriction treatment, there were no between-line differences.

PCA of the 1-Dimensional Proteomic Profiles After image analysis of the 1-dimensional electrophoresis, 42 bands were consistently resolved in samples collected from the total population (112 birds). A gel image of 2 representative samples is displayed in Figure 2, panel A. Band percentages were calculated and a correlation matrix was constructed. A PCA was performed, but only the first 2 component loadings and the first 2 component scores were reported and displayed in Figure 2, panels B and C, respectively. The first principal component explained 21.2% of the variation and the second principal component explained 12.6%, for a total of 33.8% of the variation being explained by the 2 principal components. For the extended data-set analysis, the first 5 principal components were included. These 5 principal components accounted for 54.2% of the variation. Principal component score loadings represent individual samples and show the spread of the samples in relation to the principal component loadings. Each individual turkey that was sampled is represented by a score in the plot. Samples where scores were clustered

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

BW Differences. Body weight data were imported into SAS version 9.2 (SAS Institute Inc., Cary, NC) and analyzed by a mixed model using the MIXED procedure. Significant differences were declared at a 95% confidence level. Briefly, BW was estimated by including the fixed effects of turkey line, dietary treatment, age, and all interactions. In addition, the experiment included a repeated measurement term to identify the birds that were raised together in the same battery pen. Only the 3-way interaction terms were estimated. The model is described as follows:

ses were performed: 1) used the first 2 principal component scores as depicted in the generated plots; and 2) used the first 5 principal component scores. To identify differences among treatment combinations within each time point (line × treatment), independent MANOVA were performed using a subset of the data for each time point. A set of orthogonal contrasts were used to test each pair-wise comparison within a time point. Significant differences were declared at a 95% confidence level. To evaluate the spread of the score clusters formed from the initial PCA, the mean Euclidean distances for each observation’s set of scores to the centriole of the cluster was calculated. This value represents the spread of the observations within each cluster. To obtain these values, a subset of the principal component score data for each cluster was analyzed by cluster analysis using Minitab v.15.1 (Minitab Inc., State College, PA). The mean Euclidean distance was calculated separately to the first 2 principal component scores and to the first 5 principal component scores within a cluster.

TURKEY SARCOPLASMIC PROFILE PATTERN

2831

together have more in common than samples that were further apart. The spread observed in the sample scores is related to the amount of variation that exists among them. The relationship of the individual scores to the loadings is a reflection of which variables were different among samples. Loadings that appeared in the center of the plot had less variation than those that were located closer to the edges. Based on this interpretation, the abundance of band 42 appears to be relatively homogeneous across all samples (Figure 2, panel C), whereas the abundance of band 13 appears to be more heterogeneous across all samples. In Figure 2, panel B, the sample score labeled as 7 d has a greater abundance of bands 1, 6, 13, 14, and 34, and a lesser abundance of bands 17, 21, 26, and 27. In contrast, the sample score labeled as 24 d exhibits the opposite scenario, where bands 17, 21, 26, and 27 are in greater abundance than bands 1, 6, 13, 14, and 34. Score plots were decomposed to help the visualization of developmental patterns related to line and feeding regimen. Table 1 displays the outcome of the MANOVA analyses on the principal component score data. Significant differences are consistent between the 2 separate analyses performed on the first 2 and first 5 principal component scores, except on the interaction effects between treatment × day and line × treatment × day. Table 2 displays the mean Euclidean distance of component scores to the cluster centriole, which measures the spread of the observations within each cluster. The spread observed in a cluster represents the vari-

ability among the observations, just as in the principal component loadings (explained in the previous paragraph). In addition to the mean Euclidean distances in Table 2, the outcome of the independent MANOVA pairwise comparisons is displayed. Mean Euclidean distance calculation and MANOVA are independent procedures. The decomposed score plots are displayed in Figure 3. At 7 d (Figure 3, panel A), the cluster of observation scores is located on the right side of the plot; all of the observations at 7 d had positive scores for the first principal component, with some scores above 10. Restrict-fed poults, regardless of line, exhibited a predominantly greater degree of variability than that of the ad libitum-fed poults, as observed in Table 2. The variability in 9 out of 12 comparisons between restricted- and ad libitum-fed poults was higher for the restricted group when 2 components were analyzed, and in 8 out of 12 comparisons when 5 components were analyzed. In addition, the RBC2 poults showed a greater degree of variability than that of the F line, as observed in Table 2. At 11 d (Figure 3, panel B), the cluster of observation scores moved left, closer to the center of the plot, with 11 of the observations having negative scores. This was because variations in protein abundances were reduced at 11 d compared with at 7 d, whereas the greatest degree of variability was still observed in the restrict-fed poults, similar to what was observed at 7 d (Table 2). At 14 d (Figure 3, panel C), the cluster of observation scores continued to move

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

Figure 1. Mean BW measured before kill. F corresponds to the fast-growing commercial turkey line, whereas RBC2 corresponds to the random-bred control line. Error bars represent the SD of the sample. Each point represents the mean of 4 or 5 biological replicates.

2832

Zapata et al. Table 1. P-values by multivariable ANOVA of main and interaction effects of component scores for 2 lines of turkeys1 Effect

First 2 component scores2

First 5 component scores2

0.2318 0.0306 0.0433 <0.0001 0.27 0.1802 0.0725

0.6169 <0.0001 0.0026 <0.0001 0.0991 <0.0001 0.0005

Line Feed regimen Line × feed regimen Age Line × age Feed regimen × age Line × feed regimen × age 1P-values

were obtained through Hotelling-Lawley T2 tests. analyses were performed on the first 2 and the first 5 principal component scores.

2Independent

what was observed at 14 d, with only 3 observations having positive values. At this age, however, the cluster of observation scores increased over the vertical axis. This is evident in the F-line restrict-fed poults and the RBC2 ad libitum-fed poults. The cluster of observation scores at 17 d, however, continued the trend of moving

Figure 2. A) Representative SDS-PAGE lanes of the sarcoplasmic fraction of 2 RBC2 turkeys: 1 sampled at 7 d post-hatch and the second sampled at 24 d post-hatch. B) Principal component analysis score plots of samples. Arrows in B indicate samples shown in A. C) Principal component analysis component loadings of variables (bands). Prin1 and Prin2 in C refer to first principal component and second principal component, respectively. In B and C, only the first 2 component scores are displayed.

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

slightly to the left of the plot, with only 5 observations having positive scores. The restrict-fed poults continued to show greater variation compared with that of the ad libitum-fed poults (Table 2), similar to what was observed at 7 and 11 d. At 17 d (Figure 3, panel D), the cluster of observation scores showed little change from

2833

TURKEY SARCOPLASMIC PROFILE PATTERN Table 2. Mean Euclidean distance of component scores to cluster centriole and multivariable ANOVA within each age time point of 2 lines of turkey (RBC2 and F)1,2 First 5 component scores3

7d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted 11 d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted 14 d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted 17 d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted 21 d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted 24 d   F-line, full fed   F-line, restricted fed   RBC2 line, full fed   RBC2 line, restricted

fed

1.565ab 3.160ab 2.417a 4.392b

2.219a 3.817b 3.643a 5.886a

fed

2.221 2.691 1.732 2.712

2.816 3.197 2.793 3.286

fed

1.903 2.008 1.634 2.208

2.445a 2.355b 2.448a 2.735b

fed

1.672a 2.145b 4.548ab 1.732ab

2.137a 3.727ab 6.187b 2.510a

fed

1.230 0.878 1.344 1.843

1.756a 1.756ab 1.667b 2.760ab

fed

2.210 1.602 0.681 1.099

3.458 3.341 1.363 1.529

a,bSignificant

differences are declared only in the time points where they are found and are represented by different superscripts at 95% confidence level for the cluster and not of the mean Euclidean distance. 1The mean Euclidean distance to cluster centriole measures the spread of the observations within each cluster. Each cluster corresponds to the scores that correspond to observations within the same treatment. 2Mean Euclidean distance calculation and multivariable ANOVA are independent procedures. 3Independent multivariable ANOVA were performed on the first 2 and the first 5 principal component scores.

to the left of the plot. At 21 d (Figure 3, panel E), the movement of the cluster of observation scores to the left stopped, with only 1 observation having a positive score. The amount of variation at this point was at its smallest and there was no apparent difference in variation among the 4 treatment groups (Table 2). The increase in spread over the vertical axis that was observed at 17 d was no longer present. At 24 d (Figure 3, panel F), the cluster of observation scores was similar to what was observed at 21 d, and only 2 observation scores had positive values. At this time point, the ellipses were wider than those at 21 d, but this feature was influenced because the sample size for the calculation of the ellipses was smaller, given that some of the samples died during the experiment; this notion is supported by comparing the mean Euclidean distances, which are very similar to the values at 24 d. Based on the ellipses, it is not possible to establish if the increase in variation observed at 24 d reflects a biological phenomenon. This notion is unlikely, given that the spread of the cluster of observation scores was similar to what was observed at 21 d and the mean Euclidean distances were just slightly larger (Table 2).

The spread over the vertical axis appears to be homogeneously distributed around 0 at each age, and decreased slightly with age, with the exception of the d 17 time point. This phenomenon is largely because of the variation observed in the RBC2 ad libitum-fed group on d 17 where the mean Euclidean distance for that group (Table 2) was oddly larger than those of the rest of the groups, and was also even larger than values from the previous time point. The variation was increased at 17 d along the vertical axis, and at 21 d it returned to the progression observed from 7 to 24 d.

DISCUSSION PCA of the 1-Dimensional Proteomic Profiles Figure 3 (panels A–C) shows that restrict-fed poults, regardless of line, display greater variation, represented by their mean Euclidean distances, than that of the ad libitum-fed poults. The increase in variation in restrictfed RBC2 poults suggests that they are using a larger cadre of proteins to adapt to the restricted feeding regi-

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

First 2 component scores3

Time point and treatment

2834

Zapata et al.

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

Figure 3. Decompositions of the principal component analysis score plots of samples. Each panel shows only the samples from a single time point. Each plot is coded by turkey line (F line = squares; RBC2 line = circles) and by diet (full feed = open; restricted feed = filled). The 95% confidence ellipses are displayed for the mean of each line and diet combination.

men when compared with restrict-fed F-line poults. This may be because of the ability of the RBC2 poults to express a greater number of genes in response to environmental challenges compared with that of the F-

line poults. The variation difference observed between groups (subject to different feeding regimens) indicates that there is a greater effect of feed restriction on sarcomeric protein abundance at early developmental stages.

TURKEY SARCOPLASMIC PROFILE PATTERN

to a common biological status that is independent of genetic differences or nutritional regimen, although this convergence can be delayed. In the course of evolution, natural selection has facilitated the ability of animals to optimize many biological functions. These functions have become highly conserved throughout the phylogenetic tree. The expression of some of the genes controlling biological functions may be temporally influenced by genotype or nutritional regimen, but the ultimate expression of the trait remains the same. This is best illustrated in the current study by the homogeneity of the component scores among lines and treatment groups at the d 21 time point. The second point is that there are different paths that animals can take before convergence to a similar biological status. This is suggested by the increased variability due to genotype and nutritional regimen at younger ages. This was particularly true with respect to differences in how each genetic line adapted to the restricted feeding treatment.

Conclusion The objective of the study was to use multivariate PCA to visually display differential sarcomeric electrophoretic protein profiles in 2 turkey genotypes, selected for differential growth, undergoing feed restriction during early postnatal development. The PCA successfully displayed differences in the genotype and nutritional variables at the different ages post-hatch. This method successfully illustrated the extent of participation of 1 variable compared with another over time. The comparative protein profiles of sarcoplasmic proteomic fractions from turkey breast muscles were more variable at time points before 17 d. This information will be advantageous in future studies where the objective is to characterize differences that are likely to have the greatest effect on the growth and development of poultry muscle.

ACKNOWLEDGMENTS This work was supported by the Ohio Agricultural Research and Development Center, Wooster, OH JAAP Poultry Endowment Funds to John Mark Reddish (The Ohio State University, Columbus) and Macdonald Wick (The Ohio State University, Columbus).

REFERENCES Barrett, J., P. M. Brophy, and J. V. Hamilton. 2005. Analyzing proteomic data. Int. J. Parasitol. 35:543–553. Bayyari, G. R., W. E. Huff, N. C. Rath, J. M. Balog, L. A. Newberry, J. D. Villines, J. K. Skeeles, N. B. Anthony, and K. E. Nestor. 1997. Effect of the genetic selection of turkeys for increased body weight and egg production on immune and physiological responses. Poult. Sci. 76:289–296. Christensen, V. L., W. E. Donaldson, K. E. Nestor, and J. P. McMurtry. 1999. Effect of genetics and maternal dietary iodide supplementation on turkey embryonic growth. Poult. Sci. 78:883–889. Cressman, M. D., Z. Yu, M. C. Nelson, S. J. Moeller, M. S. Lilburn, and H. N. Zerby. 2010. Interrelations between the micro-

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

The spread over the vertical axis represents the second principal component scores. The reason for the variation at a particular age is unknown, but it is mostly connected to a single treatment. The increase in variation at this time point may be because of an unknown physiological adjustment or to an external factor that was not taken into consideration and does not correspond to genotype or nutritional regimen main effects. The movement of the cluster of score patterns from right to left across the horizontal axis, corresponding to the first principal component, reflects the effect of age and developmental stage on the sarcoplasmic fraction proteomic profiles. The cluster of observation scores at the d 7 time point is located on the right side of the plot and moves progressively from right to left, stopping at approximately 17 d. This movement across the horizontal axis suggests a metabolic adaptation process that starts at hatch and is almost complete by approximately 2 wk of age. The mean Euclidean distances and the shape of the clouds are indicative of the variation in the abundance of variables or bands. A larger mean Euclidean distance and cloud area is indicative of increased variation, which may be associated with nutritional or genetic influences. In our study, the first 2 principal components were able to differentiate treatment groups within a time point (Table 2); however, this differentiation is more evident and more precise in the analysis performed using the first 5 principal component scores, where smaller P-values are observed. Some features observed in the plots and in the mean Euclidean distances are indicative of the genotype and nutritional regimen participation in the biological system. For instance, at 7 and 11 d, the mean Euclidean distances and the size of the cloud were larger for restrict-fed poults than those for ad libitum-fed poults, indicating that nutrition was delaying the convergence of protein abundances as animals accommodated to the limited nutritional plane. It is likely that the type and abundance of sarcoplasmic proteins present in the breast muscle of developing turkeys undergoes an adaptation process; however, in this study, the protein abundance of individual proteins was not determined, but is actively being pursued with further proteomic analyses by liquid chromatography/tandem mass spectrometry (LC/MS/MS). As the poults age, the variability in the abundance of sarcoplasmic proteins appears not only to change but also to become more stable and decrease. This suggests that there are fewer types of individual proteins being expressed, and results in a more biologically efficient metabolism. The shift toward a more stable state is affected by both genotype and nutritional regimen, but these treatment effects do not alter or delay the shift, at least not during the first 24 d of age, and thus, appear to be temporally driven, similar to what has been reported for myosin. The developmental patterns reflected in the sarcoplasmic protein profiles evaluated in this study illustrate 2 points. The first point is that animals converge

2835

2836

Zapata et al. parameters over thirty generations of selection for increased body weight. Poult. Sci. 79:445–452. Nestor, K. E., J. W. Anderson, R. A. Patterson, and S. G. Velleman. 2008. Genetics of growth and reproduction in the turkey. 17. Changes in genetic parameters over forty generations of selection for increased sixteen-week body weight. Poult. Sci. 87:1971–1979. Nestor, K. E., M. S. Lilburn, Y. M. Saif, J. W. Anderson, R. A. Patterson, Z. Li, and J. E. Nixon. 1999a. Influence of body weight restriction in a body-weight-selected line of turkeys on response to challenge with Pasteurella multocida. Poult. Sci. 78:1263–1267. Nestor, K. E., Y. M. Saif, J. W. Anderson, R. A. Patterson, and Z. Li. 1999b. Variation in resistance to Pasteurella multocida among turkey lines. Poult. Sci. 78:1377–1379. NRC. 1994. Nutrient Requirements of Poultry. 9th Rev. Natl. Acad. Press, Washington, DC. Supek, F., P. Peharec, M. Krsnik-Rasol, and T. Šmuc. 2008. Enhanced analytical power of SDS-PAGE using machine learning algorithms. Proteomics 8:28–31. Tidyman, W. E., L. A. Moore, and E. Bandman. 1997. Expression of fast myosin heavy chain transcripts in developing and dystrophic chicken skeletal muscle. Dev. Dyn. 208:491–504. Updike, M. S., H. N. Zerby, J. C. Sawdy, M. S. Lilburn, G. Kaletunc, and M. P. Wick. 2005. Turkey breast meat functionality differences among turkeys selected for body weight and/or breast yield. Meat Sci. 71:706–712. Zapata, I., H. N. Zerby, and M. Wick. 2009. Functional proteomic analysis predicts beef tenderness and the tenderness differential. J. Agric. Food Chem. 57:4956–4963.

Downloaded from http://ps.oxfordjournals.org/ at National Kaohsiung University of Applied Sciences on June 20, 2015

biotas in the litter and in the intestines of commercial broiler chickens. Appl. Environ. Microbiol. 76:6572–6582. doi:10.1128/ AEM.00180-10. FASS. 1998. Guide for the Care and Use of Agricultural Animals in Agricultural Research and Teaching. Fed. Anim. Sci. Soc., Savoy, IL. Graham, S. 2004. Effects of genetic selection on post-mortem meat processing. MS Thesis. The Ohio State University, Columbus. Havenstein, G. B., P. R. Ferket, J. L. Grimes, M. A. Qureshi, and K. E. Nestor. 2007. Comparison of the performance of 1966- versus 2003-type turkeys when fed representative 1966 and 2003 turkey diets: Growth rate, livability, and feed conversion. Poult. Sci. 86:232–240. Ishiwata, S., and T. Funatsu. 1985. Does actin bind to the ends of thin filaments in skeletal muscle? J. Cell Biol. 100:282–291. Liu, Y., B. G. Lyon, W. R. Windham, C. E. Lyon, and E. M. Savage. 2004. Principal component analysis of physical, color, and sensory characteristics of chicken breasts deboned at two, four, six, and twenty-four hours postmortem. Poult. Sci. 83:101–108. Lovanh, N., K. L. Cook, M. J. Rothrock, D. M. Miles, and K. Sistani. 2007. Spatial shifts in microbial population structure within poultry litter associated with physicochemical properties. Poult. Sci. 86:1840–1849. Maruyama, K., and N. Kanemaki. 1991. Myosin isoform expression in skeletal muscles of turkeys at various ages. Poult. Sci. 70:1748–1757. Nestor, K. E. 1984. Genetics of growth and reproduction in the turkey. 9. Long-term selection for increased 16-week body weight. Poult. Sci. 63:2114–2122. Nestor, K. E., J. W. Anderson, and R. A. Patterson. 2000. Genetics of growth and reproduction in the turkey. 14. Changes in genetic