Characterization of the two chicken lines
It is clear that the percentages of abdominal fat in the two lines have become very different after selecting for eight generations (Figure 1). In the eighth generation, the AFP is 2.95% in the fat chicken line and 1.55% in the lean line.
The chickens used in this study differed significantly not in body weight but in abdominal fat weight and percentage of abdominal fat (Figure 2) and were chosen on that basis. AFP in the fat line was three times that in lean line (Table 1).
Adipose tissue gene expression profile
The pattern of adipose tissue gene expression of chickens at 7 weeks was analyzed by oligonucleotide microarrays. Normalized data were used to analyze the total expressed genes. Depending on the individual bird, 13,234–16,858 probe sets were detected (bird 7-3-16 was excluded from subsequent analysis for the reason given below). Subsequently, the distribution of expression levels of genes in adipose tissue was calculated by JMP4.0 (Figure 3). The genes were ordered according to their mean expression levels; those with expression levels in the highest or lowest 1% were considered highly expressed or the converse. Some of the genes with especially high or low expression levels are named in Tables 2 and 3.
We also believe that genes described as "not expressed" can provide useful information about the function of adipose tissue and lipid metabolism in chicken. Table 4 shows some of these genes, selected using the software Affymetrix Microarray Suite 4.0. Many genes allegedly involved in lipid metabolism and obesity were not expressed in adipose tissue in the 7-week-old chickens.
Analysis of consistency within the fat and lean lines
Although there is little difference among individuals in the AFW and AFP within each line, individuals in each line may differ in hereditary molecular characters. By comparing the consistency within each line, the selective effect can be evaluated properly, and chickens that deviate too much from the norm can be excluded from subsequent screening of differentially expressed genes. This makes the analysis of differentially expressed genes more reliable and credible. The cluster analysis results showed that the fat line individuals, except chicken 7-3-16, were more consistent that the lean line ones. Chicken 7-3-16 deviated too much from the other fat line chickens (R4.
Identification of differentially expressed genes
In order to identify the differentially expressed genes, Significance Analysis of Microarrays (SAM) was performed on normalized data as described by Tusher et al. [8]. To avoid the low-variance problem in t-tests, SAM uses a statistical method similar to the t-test and estimates the false discovery rate by permutation of repeated measurements [8]. Subsequently, a two-class SAM analysis was performed on the log transformed data matrix (see Materials and Methods). A cutoff value, delta, depending on an arbitrary false positive rate, was chosen to identify genes that were significantly differentially expressed. For this analysis, a delta value of 0.8 was used (Figure 5). This led to the identification of a total of 230 differentially expressed genes: 153 were up-regulated and 77 were down-regulated in fat chickens compared to lean chickens (Figure 6). Highly differentially expressed genes were further selected by the fold change (fat/lean). These differentially expressed genes were mainly involved in lipid metabolism, energy metabolism, signal transduction, immunity and tumorigenesis. Table 5 is a summary of the most representative of these genes.
Validation of gene expression data by quantitative real-time PCR
To validate the microarray results, we performed quantitative real-time PCR for: propionyl-coenzyme A-carboxylase (PCC), similar to 1-phosphatidylinositol-4,5-bisphosphate phosphodiesterase gamma 2 (PBP2), tumor necrosis factor, alpha-induced protein 1 (TNFAIP1), fms-related tyrosine kinase 1 (FLT1), glycerol-3-phosphate dehydrogenase 2 (G3PD), low density lipoprotein-related protein 12 (LRP12), prostaglandin E receptor 3 (PER), suppression of tumorigenicity 7 (ST7), similar to endoplasmic reticulum oxidoreductin 1-Lbeta (ERO1), ataxin 3 (ATXN3), parvin, alpha (PARVA), CWF19-like 2 (CWF19), similar to Stxbp4 protein (Stxbp4), acyl-CoA oxidase (ACO), and suppressor of cytokine signaling 7(SOCS7) (Table 6). In all but one case (TNFAIP1), the real-time RT-PCR fold differences were in complete correspondence with the microarray data. Table 7 compares the microarray and real-time RT-PCR results.
Answers to all your Biology Questions
