Distinctive Genes Determine Different Intramuscular Fat and Muscle Fiber Ratios of the longissimus dorsi Muscles in Jinhua and Landrace Pigs

Meat quality is determined by properties such as carcass color, tenderness and drip loss. These properties are closely associated with meat composition, which includes the types of muscle fiber and content of intramuscular fat (IMF). Muscle fibers are the main contributors to meat mass, while IMF not only contributes to the sensory properties but also to the plethora of physical, chemical and technological properties of meat. However, little is known about the molecular mechanisms that determine meat composition in different pig breeds. In this report we show that Jinhua pigs, a Chinese breed, contains much higher levels of IMF than do Landrace pigs, a Danish breed. We analyzed global gene expression profiles in the longissimus dorsi muscles in Jinhua and Landrace breeds at the ages of 30, 90 and 150 days. Cross-comparison analysis revealed that genes that regulate fatty acid biosynthesis (e.g., fatty acid synthase and stearoyl-CoA desaturase) are expressed at higher levels in Jinhua pigs whereas those that regulate myogenesis (e.g., myogenic factor 6 and forkhead box O1) are expressed at higher levels in Landrace pigs. Among those genes which are highly expressed in Jinhua pigs at 90 days (d90), we identified a novel gene porcine FLJ36031 (pFLJ), which functions as a positive regulator of fat deposition in cultured intramuscular adipocytes. In summary, our data showed that the up-regulation of fatty acid biosynthesis regulatory genes such as pFLJ and myogenesis inhibitory genes such as myostatin in the longissimus dorsi muscles of Jinhua pigs could explain why this local breed produces meat with high levels of IMF.


Introduction
The Jinhua pig, named after Jinhua City in Zhejiag Province of eastern China, is a traditional, slow-growing breed with a high IMF content and is popular for its superior quality pork. Jinhua ham, a type of dry-cured ham produced from the meat of Jinhua pigs is the most famous brand name s in China and Jinhua ham was awarded first prize in the 1915 Panama International Merchandise Exhibition. Jinhua pigs show strong competency of oxidative metabolism and adipogenesis, which are believed to induce more satisfactory features in muscles, such as favorable meat color, marbling and flavor [1,2]. In contrast, Landrace pigs, a commercial breed of Danish origin selected over many generations for rapid growth and enhanced carcass yield, show low activities of oxidative metabolism and adipogenesis which lead to trace amounts of fat depot. As a consequence, Landrace pigs produce comparatively less flavorful pork [3][4][5]. Thus, these two pig breeds serve as ideal models to study porcine growth performance and meat quality.
Skeletal muscle is the primary abundant porcine tissue that comprises 20to 50% of total body mass among different pig breeds, and is the main tissue responsible for meat production in pigs. It is also the major metabolic tissue and contributes up to 40% of the resting metabolic rate in adult pigs [6]. Skeletal muscle is a heterogeneous tissue that is composed of four muscle fiber types including oxidative (type I and IIa) and glycolic (type IIb) fibers [7]. Muscle with a higher content of oxidative fiber contains a higher percentage of lipids, capillaries, myoglobin and mitochondria [8]. Favorable meat traits such as color, flavor and tenderness have been found to be closely associated with a higher content of oxidative fibers in muscles [9,10]. In addition, individuals with muscles that are abundant in oxidative fibers are less likely to produce pale, soft, exudative (PSE) meat. Therefore, understanding the molecular processes that govern the development and phenotypic characteristics of skeletal muscle is instrumental in the breeding of pigs with high meat quality.
Microarray technology can simultaneously examine the differential expression of a large number of genes in a given tissue [7,11] and has been widely used to compare gene expression profiles for the identification of candidate genes responsible for relevant phenotypes [12][13][14]. For example, microarray analysis showed that sexual dimorphism of adipose tissue is determined by differentially regulated sex-specific genes regardless of diet [15]. In contrast, comparison of global gene expression profiles using Affymetrix Mu11K SubB containing 6516 probe sets revealed only 49 differentially expressed genes in the quad (white muscle) and the soleus (red muscle) [16]. Based on a home-made porcine cDNA microarray carrying 5,500 cDNA clones, Bai et al. identified 115 differentially expressed genes between the psoas (red muscle) and the longissimus dorsi (white muscle) of a 22-week-old Berkshire pig [17]. Over the past decade, a tremendous amount of porcine transcriptomics data has been obtained using the pig cDNA microarray [18][19][20], while the Affymetrix porcine genome array showed particularly superior performance for swine transcriptomics [21,22]. However, reports on the comparison of global gene expression patterns in the skeletal muscles of different pig breeds at different developmental stages are lacking. In this study, a global gene expression profiling investigation was conducted to identify differentially expressed genes in longissimus dorsi muscles of Jinhua and Landrace pigs at three developmental stages using the Affymetrix GeneChipH Porcine Genome Array containing oligonucleotides representing approxi-mately 23937 transcripts from 20201 porcine genes. We found that genes involved in adipogenesis and myogenesis were differentially expressed in Jinhua and Landrace pigs. To validate the potential utility of our microarray data, we characterized the expression and function of a novel gene, pFLJ, that is one of the genes up-regulated in Jinhua pigs at the age of d90 using both drug and gene-specific small interfering RNA (siRNA) treatment approaches in cultured intramuscular adipocyte precursor cells. Our results showed that knockdown of pFLJ expression down-regulated the genes involved in fat biosynthesis and reduced fat deposition, suggesting that pFLJ is a novel regulator of adipogenesis in the muscle.

Comparison of Carcass Traits and Meat Quality Features between Jinhua and Landrace Pig Breeds
The overall appearance of a typical adult Jinhua pig is very different from that of a Landrace pig ( Figure 1A). Growth Comparison of the body weight of Jinhua and Landrace pigs at the age of d30, d60, d90, d120 and d150, respectively. Landrace pigs gained weight much faster than Jinhua pigs. Pigs were slaughtered at around the age of d30, d90 and d150 (nine individuals per stage) and d60 and d120 (three individuals per stage) for each breed. Data are presented as means 6 standard error. *P,0.05, **P,0.01. doi:10.1371/journal.pone.0053181.g001 performance, meat quality and carcass traits in Jinhua and Landrace pigs at the same age (d30, d60, d90, d120, d150, days of age) were compared. Our results showed that from the age of d30 to d150, on average, Jinhua pigs gained approximately 40 kg in weight, while Landrace pigs gained about 70 kg ( Figure 1B), demonstrating that the Jinhua were apparently growing more slowly than the Landrace. Analysis of the lean meat ratio (LMR) and loin meat area (LMA) showed that both were significantly lower in Jinhua pigs aged from d30 to d150 (Table 1). In contrast, Jinhua pigs exhibited significantly greater back fat thicknesses (BFT) and fat meat ratios (FMR) ( Table 1, P,0.01). For example at d150, BFT and FMR in Jinhua pigs were about 2-and 2.4-fold higher, respectively (BFT: 23.7 mm in Jinhua versus 12.0 mm in Landrace; FMR: 32.4% in Jinhua versus 13.3% in Landrace) ( Table 1).
It was previously reported that the Chinese Dahe pig breed displayed higher pH values (6.08) than the western crossbred Dawu sire line pig breed (5.79) 24 h postmortem [23]. A high pH value at 45 min post mortem (pH 45; 6.00-6.58) is known to correlate with a lower incidence of PSE meat [24]. We determined the pH 45 values of both Jinhua and Landrace pigs at d30, d60, d90, d120 and d150, and found that they all ranged between 6.0-and 6.5 (Table 1), with no statistically significant differences between the two breeds. These results suggested that both breeds are less likely to produce PSE meat. Meat color parameters (L*, lightness; a*, redness; b*, yellowness) are used as an index of meat quality. Analysis of the color parameters showed that there was a significant tendency for the a* value in muscle longissimus dorsi to be lower in Jinhua pigs than in Landrace pigs at the age of d150 whilst L* and b* did not differ significantly between the two breeds (Table 1). However, several reports have shown that color parameters are not an adequate indicator of meat quality when the breed has a high IMF content [25][26][27].

Jinhua Pigs have a High Content of IMF
Oil red O staining showed that the distribution pattern of fat in the longissimus dorsi muscles in Jinhua pigs was more abundant than that in Landrace pigs ( Figure 2A). Measurement of fat content revealed significant differences between Jinhua and Landrace pigs; Jinhua pigs showed a higher IMF content at all stages examined ( Figure 2B). Notably, the IMF contents in Jinhua pigs showed a steady increase from d60 (1.48%), d90 (2.25%), d120 (3.20%) to d150 (3.38%) age stages, while that in Landrace pigs remianed relatively stable from d60 (1.13%), d90 (1.28%) to d120 (1.31%) with a only slight increase at d150 (1.79%) ( Figure 2B, P,0.01). Interestingly, the IMF contents in both breeds decreased slightly from d30 to d60 ( Figure 2B; P,0.01). Our results support previous findings that Jinhua pigs have greater BFT and IMF but lower LMR and LMA than Landrace pigs at the same age [28]. These characteristics define the superior flavor of Jinhua pork [29,30].
Global Gene Expression Profiles of longissimus dorsi Muscles in Jinhua and Landrace Pigs at d30, d90, and d150 By comparing features of growth rate ( Figure 1B) and IMF content (Figure 2), we noted that the differences between Jinhua and Landrace pigs at d30, d90 and d150 three stages can be used to represent the early initiation, steady growth and maturation of myogenesis and adipogenesis in muscle, respectively. Based on this assumption, we decided to extract total RNAs from the longissimus dorsi muscles of both breeds at d30, d90, and d150 to perform microarray hybridization. Data obtained from 18 gene-chip hybridizations (nine gene-chips for each breed, three repeats for each stage) were processed according to the procedures described in Materials and Methods. We compared the global gene expression profiles of Jinhua pigs at d90 or d150 with that at d30. Our data showed that, in comparison with their expression at d30, a total of 419 differentially expressed genes were identified in longissimus dorsi muscles at d90, including 177 up-regulated genes (d90-up) and 242 down-regulated genes (d90-down) ( Table 2;  Table S1 and S2). A total of 490 differentially expressed genes were identified in longissimus dorsi muscles at d150, including 101 up-regulated (d150-up) genes and 389 down-regulated (d150down) genes (Table 2; Table S3 and S4). Clustering analysis of microarray data [31] showed that, compared with their expression   at d30, 37 genes were both d90-up and d150-up, 109 genes were d90-down and d150-down, two genes were d90-up but d150down, and six genes were d30-down but d150-up ( Table 2). In contrast, in longissimus dorsi muscles of Landrace pigs, 106 d90-up, 231 d90-down, 93 d150-up, 383 d150-down genes were identified, respectively, when compared with expression at d30 ( Table 2; Table S5, S6, S7, S8). Clustering analysis of microarray data showed that, in comparison to expression at d30, 31 genes were both d90-up and d150-up, and 64 genes were d90-down and d150-down. Interestingly, no gene was found to be d90-up but d150-down or d90-down but d150-up ( Table 2).
The fact that no or only a limited number of genes belonged to the d90-up/d150-down or d90-down/d150-up categories in both breeds suggests that the transcriptome operates sequentially to support the development of longissimus dorsi muscle during the d30 to d150 period. This provides a possible explanation for the continuous gain in muscle mass during this developmental window.
We also compared the d90-up and d90-down genes in Jinhua pigs with those of Landrace pigs. The results showed that only 0.7% of d90-up and 1.7% of d90-down genes were shared in these two breeds (Table 2). For d150-up and d150-down genes, only 3.2% of d150-up and 7.9% of d150-down genes were common to the two breeds ( Table 2). These data clearly indicates that different genes are mobilized in these two breeds to govern the development of their respective longissimus dorsi muscles.

Identification of Genes Differentially Expressed in Jinhua and Landrace Pigs during Muscle Development
The global expression profiles in longissimus dorsi muscles at d30, d90 and d150 in Jinhua pigs were compared with those in Landrace pigs at corresponding stages. A total of 375, 431 and 1195 genes were identified at d30, d90 and d150 age of stage, respectively, with at least 2.0-fold difference (P value,0.05) between two breeds (Table 3). Among these, 176, 276 and 525 genes corresponding to the stages of d30, d90 and d150 were upregulated in Jinhua pigs (Jinhua-up genes) (Table 3; Table S9, S10, S11), and 199, 155 and 670 genes corresponding to the stages of d30, d90 and d150 were down-regulated (Jinhua-down genes) ( Table 3; Table S12, S13, S14).
Among the differentially expressed genes identified by microarray in longissimus dorsi muscles of Jinhua and Landrace pigs at d90, 16 Jinhua-up genes (AY589691.1, CO993113, BF712908, CN153105, BF078710, BX924812, CF365450, NM_213785, NM_213938.1, NM_214392, BQ600160, BI399912, U83916.1, CF176622, NM_214294.1, NM_214236.1) were selected for validation by quantitative polymerase chain reaction (qPCR). Our results showed that with the exception of NM_214392 all of the selected genes were confirmed to be Jinhua-up genes ( Figure 3). Number of genes was obtained by comparing the expression profiles between Jinhua and Landrace pigs of the same age. Details are listed in Tables S10 (d30 Jinhua-up), S11 (d90 Jinhua-up), S12 (d150 Jinhua-up), S13 (d30 Jinhua-down), S14 (d90 Jinhua-down), S15 (d150, Jinhua-down). doi:10.1371/journal.pone.0053181.t003  However, we notied that, although the patterns of differential expression of the examined genes were qualitatively similar between microarray and qPCR analysis (which shows the reliability of our microarray analysis), the fold changes obtained by the two approaches differed. We reasoned that this may be due to the greater accuracy of quantitation provided by qPCR compared with microarraysor to differences in the scope of magnitude of measurement of the two techniques [32]. Adipose Deposition Related Genes are Differentially Activated in Jinhua and Landrace Pigs A high IMF ratio is considered to be the major factor that contributes to the flavor of Jinhua meat. We noted that the IMF ratio in Jinhua pigs (2.25%) was ,76% higher than that in Landrace pigs (1.28%) at d90 ( Figure 2B), suggesting that, in addition to muscle development, IMF development program in Jinhua pigs must be activated at this time-point. We analyzed the differentially expressed genes in the two breeds at d30, d90 and d150 to elucidate the relationship between differential gene expression patterns and phenotypic differences in their longissimus dorsi muscles. Table 4, Table 5, and Table 6 (for pigs at d30, d90 and d150, respectively) listed the representative differentially expressed genes known to be related to adipose deposition and muscle development based on the OMIM database of National Center for Biotechnolgy Information (NCBI) (http://www.ncbi. nlm.nih.gov/omim/) and relevant publications that described their biological function.

Transcription Factors and Signaling Molecules are Differentially Expressed in the longissimus dorsi Muscles in Jinhua and Landrace Pigs
Further analysis of the differentially expressed genes led us to identify a number of known transcription factors and signaling molecules that have not previously been reported to function in the development of longissimus dorsi muscles. Among these, we found that (bone morphogenetic protein 1 (BMP-1), regulator of G-protein signaling (RGS2) and proenkephalin (PENK) were up-regulated whereas four and a half LIM domains 3 (FHL3), F-box protein 32 (FBXO32) and a gene similar to CCAAT/enhancer-binding delta protein (LOC100153946) were down-regulated in Jinhua pigs at 30d (Table 7). Transcription regulators SWI/SNF related, matrix associated, actin dependent regulator of chromatin member 5 (SMARCA5), a gene similar to T-box 3 protein (LOC100152741) and growth arrest and DNA-damage-inducible alpha (GADD45A) were up-regulated while selenoprotein X 1 (SEPX1), homeobox protein A10 (HOXD10A) and DNA cytosine-5-methyltransferase 3 alpha (DNMT3A) were down-regulated in Jinhua pigs at d90 (Table 8). Interestingly, we noted that BMP2 and BMP receptor type 1B (BMPR1B) which mediate BMP signaling were up-regulated while secreted frizzled-related protein 4 (SFRP4) and dickkopf homolog 3 (DKK3) which mediate Wnt signaling were down-regulated at d150 ( Table 9), suggesting that key developmental signaling pathways are differentially mobilized in Jinhua and Landrace pigs. It will be of our great interest in the future to study how these transcription factors and signaling molecules control/regulate the distinct developmental events in Jinhua and Landrace pigs.

pFLJ Encodes a Novel Protein and is Highly Expressed in the longissimus dorsi Muscle of Jinhua Pigs at d90
The microarray data allowed us to search for novel genes involved in the adipogenesis process in muscles. We noted that one unknown gene corresponding to an expressed sequence tag (EST) with accession number BI184304 was expressed at a much higher level in Jinhua than in Landrace at d90. We cloned the full length cDNA corresponding to BI184304 through 59-and 39-rapid amplification of cDNA ends (RACE; data not shown) and found that this gene encodes a previously uncharacterized protein named FLJ in humans [91]. A database search revealed that FLJ is highly conserved among different species and pig FLJ (pFLJ) shares 93%, 83%, 92% and 92% homology with human, mouse, chimpanzee and rhesus monkey FLJ, respectively ( Figure 4A).
qPCR was performed to examine the expression of pFLJ in different organs/tissues in Jinhua pigs. Our results showed that pFLJ is expressed at high levels in the brain, kidney, longissimus dorsi muscle and subcutaneous fatty tissue (SF) but at a much lower level in the heart, liver, spleen and lung, demonstrating that pFLJ is differentially expressed in pigs ( Figure 4B). We then examined the expression of pFLJ in the longissimus dorsi muscles in Jinhua pigs at d30, d60, d90 and d120. Our results showed that the transcript levels of pFLJ sharply increased from d30 to d90, peaked at d90 and then decreased to a lower level at d120 (Figure 4C), thus pFLJ exhibits a dynamic expression pattern during skeletal muscle development. Because its expression levels and its dynamic expression pattern in the longissimus dorsi muscle differ between Jinhua and Landrace pigs, we wondered whether pFLJ might be involved in the process of adipogenesis. To address this question, we first established a protocol to culture intramuscular adipocyte precursor cells in vitro. These cells could be successfully induced to differentiate into adipocytes at 4 days, as judged easily by Oil Red staining (data not shown). qPCR revealed that pFLJ was expressed at a higher level in the differentiated adipocytes (data not shown). SR141716 (rimonabant, an antagonist of cannabinoid receptor 1 of mammals and commonly used as an inhibitor for fat deposition) was added to the cultured intramuscular adipocytes and the expression of pFLJ and fat contents were determined at 24-and 48-hour after treatment, respectively. Our data showed that SR14716 significantly down-regulated the transcript levels of pFLJ ( Figure 5A) and fat deposition ( Figure 5B) 48 hours after treatment.
The above data suggest a probable role of pFLJ in fat deposition. To test this supposition, three siRNAs (fs1, fs2, fs3) were designed to targets the pFLJ transcript specifically. qPCR showed that these three siRNAs efficiently knocked down the transcript levels of pFLJ in cultured intramuscular adipocytes ( Figure 6A), with fs1 showing the strongest effect at 36 hours after treatment ( Figure 6B). These cultured cells were treated with pFLJ siRNA sf1 and control siRNA NS and the contents of total triglyceride (fat) in the treated cells and free glycerol in the culture medium 36 hours after treatment were measured. We found that the total triglyceride level was significantly down-regulated ( Figure 6D), which in turn resulted in an elevation in free glycerol levels in the medium ( Figure 6E). We then examined the transcript levels of fatty acid synthase (FAS), acetyl-CoA carboxylase (ACC), adipose triglyceride lipase (ATGL) and hormone sensitive lipase (HSL) in the siRNA treated cells. FAS and ACC encode two key enzymes for the synthesis of fat while ATGL and HSL gene products are responsible for the hydrolysis of fat. We found that transcript levels of all four genes were significantly down-regulated ( Figure 6C). We therefore concluded that pFLJ is a positive regulator of fat deposition in cultured intramuscular adipocytes, probably by regulating the expression of genes that are essential for fat biosynthesis.

Conclusion
In summary, our results revealed that genes that regulate adipogenesis and myogenesis are differentially expressed in Jinhua and Landrace pigs, with Jinhua pigs expressing higher levels of adipogenesis genes and Landrace expressing higher levels of myogenesis genes. More importantly, from the microarray data, a novel gene, pFLJ, was identified as a positive factor in the regulation of fat deposition in intramuscular adipocytes. pFLJ exhibited dynamic spatial and temporal expression patterns in Jinhua pigs, with high expression in the muscle at d90. Downregulation of pFLJ by either drug treatment or siRNA-mediated gene knockdown reduced fat deposition concomitantly with the down-regulation of genes responsible for fat biosynthesis. This Table 9. List of genes encoding transcription factors and signaling molecules differentially expressed in longissimus dorsi muscle in Jinhua and Landrace pigs at d90.

Probe ID
Gene ID Gene Name Gene Symbol J/L Z score observation strongly suggests that up-regulation of pFLJ together with other factors (e.g myostatin, a myogenesis inhibitory gene) in the longissimus dorsi muscles of Jinhua pigs might play a key role in determining their high rate of IMF. Future efforts will be needed to determine the functional mechanism of pFLJ in this process. Therefore, transcriptomes for adipogenesis and myogenesis in the longissimus dorsi muscles are mobilized differentially in Jinhua and Landrace pig to produce meats with different ratios of muscle fiber to intracellular fat.

Ethics Statement
This study did not involve non-human primates. All experiments described in the study were performed in full accordance with the guidelines for animal experiments released by the National Institute of Animal Health with a permit (License No: GB/T 14925-94).

Animals
Sixty six castrated Jinhua (Jinhua II breed) and Landrace (Danish breed) pigs were raised and had ad libitum access to commercial diets (nutrients levels according to the NRC) under similar conditions during the whole experimental period. Nine individual pigs from each breed at each stages (d30, d90 and d150) and three individuals per breed at each stages (d60 and d120) were slaughtered. The longissimus dorsi muscles at the last rib were collected after exsanguinations and were subsequently divided into four portions for use in the measurement of intramuscular fat, determination of meat color, determination of pH values, and isolation of total RNA. For RNA extraction, the excised samples were directly frozen in liquid nitrogen and stored at 280uC until use.

Determination of Meat Quality
At each stage (d30, d60, d90, d120 and d150), experimental pigs were individually weighed and average bodyweights of all pigs of each breed at each stage were obtained. The BFT value was averaged from the fat thickness values measured on the first rib, last rib and the last lumbar vertebrae for each individual pig using a sliding caliper (Messschieber 0-150 mm mit Momentfeststellung Nonius 1/20 mm, Wollschlaeger). The FMR or LMR were calculated as the ratio of weight of fat meat or lean meat to the total weight of fat meat, lean meat, skin and bone of the left ham. LMA was determined by tracing its surface area at the 10th rib and calculating the area using a planimeter (Planix 5.6, Tamya Digital Planimeter, Tamaya Tecnics Inc., Tokyo, Japan). Approximately 100 g samples of longissimus dorsi muscle were used to determine IMF content using petroleum ether extraction 24 hours after slaughter [92]. One gram of muscle was collected to determine the pH 45 value. The pH 45 of the left ham was measured in the center of the longissimus dorsi muscle using a portable needletipped combination electrode (NWK binar pH-K21, CE, Germany). Color was recorded on three 10 mm diameter spots from each longissimus dorsi muscle eye rib surface within 2 hours following loin slicing. Indicators of lightness (L*), redness (a*), and yellowness (b*) were recorded in triplicates by a Minolta chromameter (CR-300, Minolta Camera Co., Japan) on a freshly cut surface 45 minutes postmortem, and the average value of the three spots was used.

Extraction of RNA Samples from Muscles for Microarray Hybridization
Approximately 100 mg of frozen muscle tissues were homogenized in liquid nitrogen using a mortar and pestle under RNasefree conditions. Total RNA was extracted from the samples with Trizol-Phenol reagent (Invitrogen) according to the manufacturer's protocols. Residue DNA was removed with DNaseI (37uC, 30 mins) followed by purification of total RNA with RNeasy Mini kit (Qiagen). RNA was quantified using NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies) at 260 and 280 nm and the integrity of RNA was determined by denaturing agarose gel electrophoresis. The quality of total RNA was further assessed using an Agilent Bioanalyzer 2100 (Agilent Technologies) based on the RNA integrity number (RIN) value. All samples used for microarray analyses had an RIN value above 8.

Microarray Hybridization
Total RNA from a total of 54 pigs at d30, d90 and d150 of age stage (nine pigs for each breed at each stage) was extracted. RNA samples from three pigs of the same breed at the same age stage were pooled as one sample for one gene-chip hybridization. Microarray data from three samples for each breed at each stage were obtained for data analysis. A total of 18 microarrays were used in the experiment, corresponding to the 18 pooled RNA samples from longissimus dorsi muscles. The GeneChip Porcine Genome Array (Affymetrix, Santa Clara, CA) contains 23937 probes sets interrogating 23256 transcripts, representing 20201 genes. RNA labeling and Affymetrix Gene Chip microarray hybridization were conducted according to the Affymetrix Expression Analysis Technical Manual. Array scanning and data extraction were carried out following procedures recommended by Affymetrix.

Microarray Data Analysis
To quantify the intensities from the same probe sets on different arrays, these were scaled so that the median intensities for all arrays were the same. We then calculated the average intensity for each probe in all replicate arrays and this mean intensity was used for downstream analysis. When comparing gene expression between different breeds at the same time-point and in the same tissue, Lowess intensity dependent normalization was performed for each array pair. Z-scores were then calculated as described previously [93] and Z-scores $2 or #2 was used as the cut-off value for selection of up-or down-regulated genes. Hierarchical and K-means clustering of differentially expressed genes was done using Cluster 2.10 and viewed in TreeView 1.50 from Eisen Lab (http://rana.lbl.gov/EisenSoftware.htm).
qPCR Primer sequences, melting temperatures and expected product sizes for the genes analyzed are shown in Additional file 15 (Table S15). The sizes of the PCR products were confirmed using agarose gel electrophoresis (1.8%). The specificity of the PCR products was judged based on a single peak observed in dissociation/melting curves. All RNA samples prepared for gene-chip hybridization were also used in qPCR. qPCR was performed using SYBR green I nucleic acid dye on an BIO-RAD CFX96 Real-Time PCR System (BIO-RAD, Foster City, CA, USA) to quantify the target genes expression levels. Data are expressed as the ratio between expression of the target gene and that of the housekeeping gene 18s rRNA. All qPCR reactions followed this thermal profile: after an initial denaturation at 94uC for 2 minutes, amplification was performed with 40 cycles of 94uC for 30s and annealing for 40 s at temperatures specific for each target genes. For each sample, reactions were set up in triplicate to ensure the reproducibility of the results. At the end of the PCR run, melting curves were generated and analyzed to confirm non-specific amplification, and the mean value of each triplicate was used for further calculations. To calculate the mRNA expression of selective genes, the DCt values was used for detection of their mRNA related to internal control 18s rRNA expression using the 2 2ggCt method [94].

Cloning of the pFLJ Gene
To obtain the full-length cDNA sequence of pFLJ, RACE technology was carried out to clone the 59-ends of pFLJ by using the SMARTTM RACE cDNA Amplification Kit and GeneRacer Kit (Invitrogen Biotechnology Co. Ltd., Shanghai, China). Briefly, for 59-RACE, 59 phosphates and the 59 cap structure were removed from the total RNA from porcine tissues, the GeneRacer RNA Oligo sequence (59-CGACUG-GAGCACGAGGACACUGACAUGGACUGAAGGAGUAGA-AA-39) to the 59 end of the prepared mRNA was ligated and the 59 RACE cDNA template was then obtained by reversetranscribing the ligated mRNA according to the manufacturer's instructions. Four steps were required to obtain the full length of pFLJ36031 cDNA. The first reaction of PCR was performed using a combination of sm-FLG-R1 (59-GCCACCAATGAC-CAAAGGCACTTGGATAA-39) and 10*UPM using the 59 RACE cDNA template. The PCR condition was as follows: 94uC for 2 min, 5 cycles of 94uC for 30 s and 72uC for 1.5 min, 5 cycles of 94uC for 30 s and 70uC for 1.5 min, 25 Data are presented as means 6 standard error. *: P,0.05, **: P,0.01 (C and D) Measurement of total triglyceride (as before) in the cultured adipocytes or free glycerol (the free glycerol release was normalized to total cellular protein and expressed relative to the control group) in the culture medium 36 hours after treatment with fs1 siRNA. ab means every two columns with different letters are significantly different (P,0.05). doi:10.1371/journal.pone.0053181.g006 cycles of 94uC for 30 s, 65uC for 30 s and 68uC 3.0 min. Then the product was further identified using another primer (sm-FLG-R2:59-GCCTGATCAACGATTCCTGTGGTCTTCA-39) that is located on the downstream of sm-FLG-R1. The PCR condition used was: 94uC 2 min, 30 cycles: 94uC 30 s 66uC 30 s and 68uC 1.5 min. The gene-specific primer sm-FLG-R1 was designed based on the pFLJ EST available in GenBank. The resulting PCR product obtained from this step was isolated, cloned, and sequenced. The three subsequent 59-RACE products were gel-purified, cloned, and sequenced. By ligation of the four overlapping cDNA fragments, full-length pFLJ cDNA was obtained. Primer pairs used for qPCR were: sense: 59-cca cct ttc cca cca ttc g-39; antisense: 59-agc ctc acc acg ggt tcc ag-39.

In vitro Culture of Intramuscular Adipocyte Precursor Cells and Induction of Adipocytes
For in vitro culture of intramuscular adipocyte precursor cells, D (Duroc) 6L (Landrace)6Y (Yorkshire) pigs from d5 to d7 of age were overdosed with sodium thiopental and exsanguinated. The longissimus dorsi muscle was removed and porcine pre-adipocytes were prepared by previously published methods [95,96]. Briefly, longissimus dorsi muscle tissue was cut with scissors into approx-imately1 mm sections under sterile condition and digested with collagenase type II for 45 hours, at 37uC in a 120r/min shaking water bath. The digested material collected was first centrifuged at100 g for1 min, and the resulting floating adipocytes were collected in Dulbecco's Modified Eagle Medium (DMEM) at 37uC. The number of intramuscular pre-adipocytes isolated in suspension was determined as described previously. The preadipocytes were seeded on six-well (35-mm) tissue culture plates in complete media (DMEM/F12+10% fetal bovine serum (FBS)+100 Upenicillin+100 Ustreptomycin) and cultured at 37uC under a humidified atmosphere of 95% air and 5% carbon dioxide according to previous study [42].
Intramuscular preadipocytes were induced to differentiate into intramuscular adipocytes when the cells were completely fused and were then treated with a final concentration of 0.5 mmol/L 3isobutyl-1-methylxanthine (IBMX), 1mmol/L dexamethasone (DEX) and 1.7mmol/L insulin of complete medium. The culture medium was changed to complete medium containing a final concentration of 10 mg/L insulin after 48 hours.

Statistical Analysis
All experimental data of comparisons between two pig breeds were analyzed using one-way analysis of variance (ANOVA, Statistical Product and Service Solutions (SPSS) 16.0). Data are represented as means6standard error; *P,0.05 and **P,0.01 displayed here indicate statistically significant difference.

Supporting Information
Table S1 177 genes upregulated in longissium dorsi muscles of jinhua pig at d90 compared with that at d30 age stage (Jinhua-d90-LD-up vs d30).