- Open Access
Association between plasma metabolites and gene expression profiles in five porcine endocrine tissues
© Yang et al; licensee BioMed Central Ltd. 2011
- Received: 11 March 2011
- Accepted: 25 July 2011
- Published: 25 July 2011
Endocrine tissues play a fundamental role in maintaining homeostasis of plasma metabolites such as non-esterified fatty acids and glucose, the levels of which reflect the energy balance or the health status of animals. However, the relationship between the transcriptome of endocrine tissues and plasma metabolites has been poorly studied.
We determined the blood levels of 12 plasma metabolites in 27 pigs belonging to five breeds, each breed consisting of both females and males. The transcriptome of five endocrine tissues i.e. hypothalamus, adenohypophysis, thyroid gland, gonads and backfat tissues from 16 out of the 27 pigs was also determined. Sex and breed effects on the 12 plasma metabolites were investigated and associations between genes expressed in the five endocrine tissues and the 12 plasma metabolites measured were analyzed. A probeset was defined as a quantitative trait transcript (QTT) when its association with a particular metabolic trait achieved a nominal P value < 0.01.
A larger than expected number of QTT was found for non-esterified fatty acids and alanine aminotransferase in at least two tissues. The associations were highly tissue-specific. The QTT within the tissues were divided into co-expression network modules enriched for genes in Kyoto Encyclopedia of Genes and Genomes or gene ontology categories that are related to the physiological functions of the corresponding tissues. We also explored a multi-tissue co-expression network using QTT for non-esterified fatty acids from the five tissues and found that a module, enriched in hypothalamus QTT, was positioned at the centre of the entire multi-tissue network.
These results emphasize the relationships between endocrine tissues and plasma metabolites in terms of gene expression. Highly tissue-specific association patterns suggest that candidate genes or gene pathways should be investigated in the context of specific tissues.
- High Density Lipoprotein Cholesterol
- Plasma Metabolite
- Metabolic Trait
- Endocrine Tissue
- Normalize Enrichment Score
In recent years, high-throughput genomic technologies have accelerated the discovery of new causal mutations and made the study of biological systems more accessible than ever. This is true not only in humans and model organisms but also in agriculturally important species like the pig. One major interest in the study of livestock species is that the strong selection pressure applied in breeding programs has resulted in breeds that are phenotypically extreme for many traits. In addition, such selection has indirectly acted on the transcriptome and the metabolome, but the resulting effects are much less studied, not to say understood, than those on external phenotypes like growth or fat deposition.
In humans and other animal species, the blood levels of molecules related to lipid, glucose and protein metabolism, such as non-esterified fatty acids, triglyceride, glucose and alanine aminotransferase (ALT), reflect nutritional and disease status. In livestock species, the abundance of plasma metabolites can be associated with agriculturally important traits like growth and fatness . Among the major livestock species, pig is a good model for human diseases such as atherosclerosis . Genetic mapping studies have identified several genetic loci affecting blood metabolites in both human and pig populations [3, 4]. Ideally, the functions of genes need to be defined in the context of relevant tissues and gene expression networks. Most of the studies that combine gene expression network and data on plasma metabolites have been primarily carried out on liver and adipose tissues [5, 6]. However, endocrine glands, by secreting hormones, also play a pivotal role in maintaining the homeostasis of plasma metabolites, either directly or indirectly. Despite the importance of these tissues, the relationship between endocrine transcriptome and plasma metabolites is not well known. In addition, most existing analyses have considered tissues separately although complex traits like obesity or metabolite blood levels involve molecular networks both within and between multiple tissues.
In the work reported here, we have analyzed the association between the transcriptome of five endocrine tissues (hypothalamus, adenohypophysis, thyroid gland, gonad and fat tissue) and 12 plasma metabolites in pig. Since the study was carried out on pigs belonging to different breeds but managed and sacrificed simultaneously, we could also investigate the existence of any genetic (breed) effect on the metabolites analyzed. The plasma metabolites studied here play a fundamental role in the basal metabolism (glucose, cholesterol, triglyceride and non-esterified fatty acids, alanine aminotransferase), or the inflammatory response (haptoglobin, pig major acute phase protein). The term "quantitative trait transcript" or QTT refers to a probeset, the expression of which is significantly associated (P < 0.01) with a particular metabolic trait. Gene co-expression networks, were inferred both for each tissue separately and for all tissues together. We conclude that using a multi-tissue network provides key relevant information to understand the underlying regulation of the metabolites studied.
Animals and sample collection
Animal management and tissue collection procedures have been detailed elsewhere . Briefly, 27 pigs from five breeds, Large White (N = 6), Landrace (N = 5), Duroc (N = 5), a Sino-European hybrid line (N = 5) and Iberian (N = 6), were bought from three breeding companies after weaning. All pigs were housed together in the university experimental farms and fed the same diet for two months. At 80 to 89 days of age and after 24 hours fasting, pigs were euthanized and sacrificed for blood and tissue sampling. All procedures were approved by the Ethical and Animal Welfare committee of the Universitat Autònoma de Barcelona (Spain).
Twelve plasma metabolites were measured in the 27 pigs. Briefly, after collecting and coagulating blood samples at room temperature, serum was separated from clots by centrifugation at 3000 rpm at 4°C for 20 min and stored at -80°C until use. Plasma metabolite concentrations were measured with the following methods: hexokinase assay for glucose, Ranbut assay (Randox Laboratories Ltd., UK) for 3-hydroxybutyrate, NEFA-C reagent (Wako Chemicals GmbH, Germany) for non-esterified fatty acids (NEFA), CHOD-PAP-method for cholesterol, immuno-inhibition method for high density lipoprotein cholesterol (HDL-C), selective protection method for low density lipoprotein cholesterol (LDL-C), GPO-PAP method for triglyceride, Biuret method for total protein and, methods recommended by IFCC (International Federation of Clinical Chemistry) for alanine aminotransferase (ALT) and alkaline phosphatase (ALP). Haptoglobin was assayed with the Phase Haptoglobin kit (colorimetric assay based on binding of haptoglobin to hemoglobin, Tridelta Ltd, Ireland) and pig major acute phase protein (PigMAP) levels with an ELISA kit (PigCHAMP ProEuropa, Segovia, Spain). All the assays were performed with an Olympus AU400 analyzer according to the manufacturer's recommendations.
We used the GeneChip® Porcine Genome Array from Affymetrix (Santa Clara CA) to profile the transcriptome of five endocrine tissues: hypothalamus (HYPO), adenohypophysis (AHYP), thyroid gland (THYG), gonads (GONA) from both male and female pigs, and backfat tissue (FATB) in 16 (four Large White, four Duroc, four Iberian and four from the Sino-European hybrid line) of the 27 pigs. Each breed consisted of two males and two females, except for the hybrid line with three males and one female . Total RNA was extracted from 100 mg of tissue and RNA samples were cleaned, quantified, and adjusted to 500-1000 ng/μl. Five μg of total RNA were used to synthesize cDNA. Then, the 80 microarrays corresponding to 16 animals × five tissues were hybridized and scanned to generate signal intensities which were converted to CEL files by the GeneChip Operating Software (GCOS). All CEL files were adjusted for background noise and normalized using the GCRMA procedure  and the data was then used for subsequent analysis. The transcriptome data are deposited in the Gene Expression Omnibus (GEO) database under accession number [GEO:GSE14739].
Data processing and analysis
where y is a vector of the studied metabolite measures.
where probeset i is defined as a quantitative trait transcript (QTT) if its association with a particular biochemical trait achieves a nominal P value < 0.01. Since both breed and sex were adjusted in the regression analysis, the detected QTT for a particular metabolite represent general transcriptional effects in both breed and sex. The analysis were implemented using the GLM function in R . The False Discovery Rates (FDR) of the associations were determined by permuting the labels of the phenotypes for 20 iterations, while preserving the correlation structure of the transcriptome.
Gene set enrichment analysis
A gene set enrichment analysis (GSEA) was implemented using R scripts downloaded from http://www.broadinstitute.org/gsea/ with a few modifications. In this analysis, the average value across probesets was used as the expression value of that gene in each individual when a gene was represented by more than one probeset. This reduced the 24,123 probesets to 18,017 unique genes. For each metabolic trait, we ranked the 18,017 genes according to their partial correlations with the metabolic trait under study (conditional on sex and breed). Then, an enrichment score measuring the extent to which a predefined set of genes (e.g., genes in a specific KEGG for Kyoto Encyclopedia of Genes and Genomes category) clustered at the top or the bottom of the ranks is calculated for each gene set. The normalized enrichment scores were used to measure the strength of the association between gene sets and the metabolic trait. The significance and FDR of the associations were determined by 1000 permutations .
Weighted gene co-expression network analysis
The gene expression data were corrected for sex and breed effects, and corresponding residuals were used to build up a weighted gene co-expression network using R package weighted gene co-expression network analysis (WGCNA) [11, 12]. Briefly, a Pearson correlation matrix was first obtained and then transformed into an adjacency matrix A using a power function a ij = |r ij | β , where |r ij | is the absolute value of Pearson correlation coefficients between probeset i and probeset j, a ij is the element in A. The network connectivity (K) of probeset i is defined as where index j corresponds to all probesets other than probeset i in the network, N is the overall number of transcripts studied . The parameter β is chosen so that the connectivity distribution approximates a scale-free criterion, P(K) = K -r . The adjacency matrix was further transformed into a distance matrix through topological overlap-based dissimilarity measures; finally a dynamic clustering procedure was applied on the distance matrix to divide the entire co-expression network into multiple modules . Similarly, the intramodular connectivity probeset i was defined as , where index j indicates all probesets other than probeset i in a specific module of size N m .
We also introduced a standardized inter-tissue connectivity of probeset i: , which measures the connection strength for a probeset i to probesets in external tissues, here index l indicates all the N ot probesets in tissues other than the tissue to which probeset i corresponds. The strength of connection between a pair of tissues with regard to gene expression is defined as , where i and j correspond to probesets in tissue 1 and tissue 2, and N 1 and N 2 are the number of probesets in tissue 1 and tissue 2, respectively.
Gene ontology (GO) and KEGG pathway enrichment analysis
The porcine Affymetrix probeset identifiers were converted into their human orthologs using the latest annotation file version (2010) from . The gene category enrichment analyses were conducted using the Database for Annotation, Visualization and Integrated Discovery (DAVID) web-accessible program .
Breed and sex differences for metabolite traits
Characteristics and statistics of the 12 plasma metabolites analyzed in this study
energy source of brain, rise when blood glucose is low
starvation, insulin resistance and blood pressure
progression of atherosclerosis, diet
inverse predictor of cardiovascular disease
high level Associated with cardiovascular disease
atherosclerosis, heart disease and stroke, diet
Total protein (g/L)
reflects albumin concentration, infection, inflammation.
rises dramatically in acute liver damage
rises with large bile duct obstruction, liver disease
infection, inflammatory and pathological lesion, stress
infection, inflammatory and pathological lesion, stress
Differences in metabolite levels among breeds were also visualized with a dendrogram, these differences being defined as 1 - r, where r is the correlation coefficient between standardized average values of 12 metabolites in any two breeds. Note that a perfect positive correlation corresponds to 0, no correlation to 1 and a perfect negative correlation to 2 on the y axis (Figure 1b). To facilitate the comparison with the dendrograms built with gene expression data, only the 16 animals with transcriptome data were used. As shown in Figure 1b, the Iberian and Large White breeds were within the same clade, whereas the Duroc breed and the Sino-European hybrids clustered together in a distinct clade. The height of these two clades was approximately equal to 1, meaning that the metabolite levels between Iberian and Large White pigs, and between Duroc and Sino-European pigs were uncorrelated, whereas the total height of the tree was ~ 1.6, suggesting a negative correlation between clades. Notably, we observed similar patterns in dendrograms constructed using a Bayesian standardized measure of the breed's gene expression levels  in adenohypophysis, thyroid gland, backfat tissue, hypothalamus, and female gonad (Figure 1c-e).
Association between transcriptome and plasma metabolites
Number of QTT for each plasma metabolite measured in five tissues
Non-esterified fatty acids
Tissue pairs with a significant number of overlapping QTT
Bonferroni P value
Gene co-expression networks
A gene co-expression network is a representation of how transcripts are correlated. Genes within the same biological pathway can be highly correlated and therefore grouped into the same module. Using weighted gene co-expression network analysis, the QTT for each of the 12 metabolic traits in each of the five tissues were clustered into one to four modules. Because the networks were constructed using probesets separately for each tissue, we refer to these networks as single-tissue networks. Furthermore, we examined the biological significance of these modules by gene ontology (GO) categories (including biological processes, molecular function and cellular component) and KEGG pathways enrichment analysis. The enrichment of these gene categories was assessed by p values corrected by the Benjamini and Hochberg approach .
Enrichment of gene categories in different tissue modules for NEFA, HDL-C, triglyceride, glucose and ALT levels
RNA binding and splicing
biosynthesis of unsaturated fatty acid
regulations of protein, nucleotide metabolism
T cell proliferation
muscle and skeletal development
protein transport and localization
calcium ion binding
learning and memory
RNA processing; ribosome
monosaccharide catabolic process
leukocyte transendothelial migration
nucleoside triphosphate metabolic process
blood vessel development
regulation of cell motion
polysaccharide and heparin binding
ECM receptor interaction
muscle, heart and bone development
regulation of transcription and metabolic processes
response to wounding
learning and memory
Highly connected (hub) nodes constitute the backbones of a network structure. In Figure 3d, we show a subset of the entire network using the top 10% probesets with the highest intra-modular connectivity (hub nodes). Several interesting observations can be made. All hub nodes in module 1 corresponded to adenohypophysis, while all hub nodes in modules 2 and 4 corresponded to gonad, these modules possibly reflecting biological processes that operate within tissues. In contrast, hub nodes in module 3 corresponded to four tissues including hypothalamus, thyroid, adenohypophysis and backfat, suggesting that the genes in this module could be part of gene regulation pathways that are involved in communications between tissues. Notice that 64% (73/114) of the hub genes in module 3 corresponded to hypothalamus, which is regarded as an organ integrating information from the body's nutritional and hormonal signals. Both positive and negative correlations among hub nodes were present in module 3, indicating the existence of feedback signaling. In comparison, only positive correlations among probesets within the three other modules were observed. There are many more links between module 1 and module 3 than between any other pair of modules. Many of these are links between hypothalamus and adenohypophysis genes. Interestingly, hormone secretion in the adenohypophysis is directly regulated by neurons in the hypothalamus. Thus, these observations emphasize the central role of the hypothalamus with regard to gene regulation networks.
Plasma metabolite levels are main indicators of endocrine status, including health status, and are potential predictors of performance. In this study, a survey of 12 plasma metabolites showed that six metabolites, including total protein, NEFA, ALP, LDL-C, haptoglobin and PigMAP are affected by breed (P < 0.05) and therefore have a partial genetic cause. The Iberian pig, which is fatter and grows more slowly than commercial pig breeds, has the highest average levels of total protein and PigMAP, but the lowest level of ALP. Interestingly, ALP is reported to be associated with body weight in pigs . The Sino-European hybrid pigs have lower haptoglobin and PigMAP average levels which are positively associated with inflammatory processes. This suggests that the Sino-European hybrid pigs could have a weaker inflammatory response as compared to e.g., Duroc and Landrace breeds (Figure 1a). Notably, we observed a similar pattern of correlation among breeds in terms of both the levels of the 12 metabolites and the transcriptome in multiple tissues (Figure 1b-e).
The endocrine glands play important roles in maintaining homeostasis of metabolites in blood. Here, we report an association analysis between gene expression profiles in five endocrine tissues and plasma metabolites in pigs. The associations were found to be highly tissue-specific, as suggested by the limited overlap of QTT and biological pathways in the five tissues for all the metabolites. The QTT for NEFA, ALT, HDL-C, triglyceride and glucose within each tissue were grouped into biologically meaningful sub-networks. Furthermore, we constructed a multiple-tissue network using QTT from the five tissues for NEFA.
Overall, the FDR of the associations between probesets and metabolites was high at the current significance threshold (P < 0.01) and a similar high FDR was also observed at a stricter threshold (P < 0.001). This is likely due to the limited size of the sample (N = 16). Yet, we did find a significant increase in the number of QTT for NEFA and ALT, and the QTT within tissues were grouped into biologically meaningful modules (detailed below).
The limited overlap between QTT and gene pathways across tissues suggests that the associations between endocrine transcriptome and biochemical traits were highly tissue-specific. This is in agreement with our previous analyses of the data as well  and with the literature in general. For instance, Yang et al.  have reported a minimal overlap and very different functional categories of sexually dimorphic genes in brain, liver, adipose and muscle of mice. Therefore, candidate genes or gene pathways e.g., obtained from genome-wide association studies should be investigated in the context of specific tissues.
Single tissue network
The most significant observations regarding QTT number concerned NEFA. NEFA derive from the hydrolysis of triglycerides in adipose tissue or lipoproteins, circulate in the blood and serve as source of energy (especially for heart and muscle) and cellular signaling messengers. In the backfat module, we found that genes involved in the biosynthesis of unsaturated fatty acids (such as ELOVL6, ACOT4, ACOT7, HSD17B12, PECR and SCD) were negatively correlated with NEFA, suggesting that the synthesis of unsaturated fatty acids was repressed in animals with higher plasma NEFA levels. Moreover, other genes involved in fatty acid and lipid metabolisms (such as DECR1, ACADL, ACOX2, DCI, ECHDC2, FABP3, FASN, LIPA, PRDX6, ENPP2, DDHD1, DGAT2 and SCP2) were also found negatively correlated with NEFA in this module. The hypothalamus module for NEFA was enriched for genes related to synapses, learning and memory. Many genes participating in protein transport and localization processes like SENP1, CDK5, SYNGR1, SNAP23, RIMS1 and YWHAZ are also active at synapses. Synaptic plasticity in the hypothalamus is known to be associated with nutritional state . In the adenohypophysis module, genes involved in calcium ion binding, protein transport and localization, neuron projection and in the presynaptic membrane were overrepresented. The importance of calcium-dependent electrical activity in adenohypophysis cells has been reviewed, e.g., by . Both in vitro  and in vivo  experiments have shown that changing NEFA concentrations can alter pituitary hormone secretion in pigs. Both in humans and dog, it was shown that the plasma NEFA level increases after administration of growth hormone , NEFA in turn can block growth hormone secretion . Thus, in general, we observe that enriched functional categories often have a physiological interpretation.
For ALT, the most relevant tissues in this analysis are the thyroid and gonad (Tables 2 and 4). The observed large number (3322) of QTT and wide range of functional categories in thyroid suggest a close relationship between thyroid function and plasma ALT levels. It is well known that ALT blood levels reflect the liver condition since clinical links between the thyroid and liver are well documented. Liver metabolizes the thyroid hormone, which in turn influences the liver function and thyroid disorders are often associated with an elevation of ATL . In the gonad module, we checked the genes in the enriched functional categories using DAVID online tools http://david.abcc.ncifcrf.gov/, and found that many genes (CLDN3, CLDN4, PTK2B, EPAS, CDH1, CDH2, TYMP, TGFA, WT1, CTGF, FN1 and ITGB3) related to cell adhesion or migration were associated to ovarian tumors. Moorthy et al. (2005) reported that administration of gonadal hormones like estradiol and progesterone decreased ALT levels in heart, liver, kidney and uterus in naturally menopausal rats .
For HDL-C, the backfat module, was slightly enriched for apolipoprotein genes including APOB, APOA4, APOC3, APOC4 and APOH (PBenjamini = 0.061). This observation is unexpected, since no evidence was found to support the synthesis of these apolipoproteins in adipose tissue. Both thyroid gland and hypothalamus modules contain a group of genes participating in mRNA processing specifically mRNA splicing. Alternative pre-mRNA splicing plays an important role in the control of neuronal development and function . Thyroid hormones and their receptors have been shown to stimulate reverse cholesterol transport in animal models .
To explore the connections between tissues at the gene expression level, we built a co-expression network containing all the QTT for NEFA from the five tissues (Figure 3). Module 3, in which hypothalamus genes are overrepresented, appears to be particularly interesting. The top 10% most connected genes in this module are from four different tissues and might constitute core regulation pathways involved in communication between tissues. Additionally, we have also shown that genes have a significantly higher average inter-tissue connectivity in the hypothalamus than in other tissues (Figure 3b). These observations emphasize the central role of hypothalamus genes in the multiple-tissue co-expression network. Dobrin et al. (2009) constructed inter-tissue co-expression networks between hypothalamus, liver and adipose tissue. Their results also suggested the hypothalamus as the controlling tissue since asymmetric connectivity was more common in the hypothalamus than in other tissues. e.g., the most connected hypothalamus gene, Aqp5 was linked to 169 adipose genes, while adipose gene Aqp5 was only linked to two hypothalamus genes. Interestingly, the hypothalamus is known as an organ that integrates and responds to signals from peripheral tissues [28, 29].
More links were found in hub genes between modules 1 and 3 than between any other modules (Figure 3d), suggesting that the genes in these two modules act in a more coordinate fashion. Several hypothalamus genes (FAM69B, NPTXR, RUNDC3A, N4BP2L2, KIAA1429, SNURF and KCTD20) and a backfat gene (RUNDC3B) in module 3 were highly connected to hub genes in module 1. Furthermore, we examined the hub genes in module 3 (Additional file 5: Table S5) using DAVID online tools , and highlighted the genes associated with functions in corresponding tissues. We found genes in the hypothalamus that were related to the differentiation and development of the central nervous system (ATP7A, CDK5, HPRT1 and SS18L1) and to protein transport and localization (ARFIP1, RAB6B, SENP2, C11orf2, PACS1, RIMS1 and TNKS). Most of these genes are relevant to neuron function or energy balance e.g., CDK5 is a member of the cyclin dependent kinase family, and serves as an essential modulator of synaptic function and plasticity . RAB6B is a GTPase predominantly expressed in brain that has been suggested to participate in retrograde transport of cargo in neuronal cells . This gene was also up-regulated in the brain of mice fed with omega 3 polyunsaturated fatty acid enriched diet . TNKS is a Golgi associated poly-ADP-ribose polymerase gene, abundantly expressed in brain. TNKS-deficient mice show an increase in energy expenditure, fatty acid oxidation and insulin simulated glucose utilization . Two transcripts in the backfat module correspond to UGP2, which is involved in the synthesis of UDP-glucose, the precursor of glycogen in liver and muscle tissue, and of lactose in lactating mammary gland. Among the adenohypophysis genes, FTO, RHOB and ELP2 are related to the function of the adenohypophysis. FTO is a well studied gene that is abundantly expressed in the hypothalamus and adenohypophysis and related to food intake and obesity [34, 35]. RHOB is a GTP-binding protein involved in vesicular trafficking in anterior pituitary cells . ELP2 is an isoform of steroidogenic factor 1 (SF1), and plays an important role on pituitary gonadotrope function.
Our results suggest a partial hereditary basis for some metabolite levels, like total protein, NEFA, ALP, LDL-C, haptoglobin and PigMAP. Single-tissue gene co-expression networks were composed of highly connected modules associated with metabolites (especially NEFA) that had a biologically meaningful role. These networks were, in general, tissue-specific. Finally, the multiple-tissue network emphasized the central role played by the hypothalamus.
We thank all the people involved in tissue collection for this experiment, in particular, M. López-Béjar, A.L. Ferraz and L. Fernandes. BY is funded by a scholarship (File No. 2008836039) from the China Scholarship Council (CSC). The work was funded by grants AGL2007-65563-C02/GAN and AGL2010-14822/GAN to MPE and by a Consolider grant from Spanish Ministry of Research, CSD2007-00036 "Centre for Research in Agrigenomics".
- Pond WG, Su DR, Mersmann HJ: Divergent concentrations of plasma metabolites in swine selected for seven generations for high or low plasma total cholesterol. J Anim Sci. 1997, 75 (2): 311-316.PubMedGoogle Scholar
- Turk JR, Laughlin MH: Physical activity and atherosclerosis: which animal model?. Can J Appl Physiol. 2004, 29 (5): 657-683. 10.1139/h04-042.View ArticlePubMedGoogle Scholar
- Sabatti C, Service SK, Hartikainen AL, Pouta A, Ripatti S, Brodsky J, Jones CG, Zaitlen NA, Varilo T, Kaakinen M, Sovio U, Ruokonen A, Laitinen J, Jakkula E, Coin L, Hoggart C, Collins A, Turunen H, Gabriel S, Elliot P, McCarthy MI, Daly MJ, Järvelin MR, Freimer NB, Peltonen L: Genome-wide association analysis of metabolic traits in a birth cohort from a founder population. Nat Genet. 2009, 41 (1): 35-46. 10.1038/ng.271.PubMed CentralView ArticlePubMedGoogle Scholar
- Chen R, Ren J, Li W, Huang X, Yan X, Yang B, Zhao Y, Guo Y, Mao H, Huang L: A genome-wide scan for quantitative trait loci affecting serum glucose and lipids in a White Duroc x Erhualian intercross F(2) population. Mamm Genome. 2009, 20 (6): 386-392. 10.1007/s00335-009-9190-9.View ArticlePubMedGoogle Scholar
- Ferrara CT, Wang P, Neto EC, Stevens RD, Bain JR, Wenner BR, Ilkayeva OR, Keller MP, Blasiole DA, Kendziorski C, Yandell BS, Newgard CB, Attie AD: Genetic networks of liver metabolism revealed by integration of metabolic and transcriptional profiling. PLoS Genet. 2008, 4 (3): e1000034-10.1371/journal.pgen.1000034.PubMed CentralView ArticlePubMedGoogle Scholar
- Wolfs MG, Rensen SS, Bruin-Van Dijk EJ, Verdam FJ, Greve JW, Sanjabi B, Bruinenberg M, Wijmenga C, van Haeften TW, Buurman WA, Franke L, Hofker MH: Co-expressed immune and metabolic genes in visceral and subcutaneous adipose tissue from severely obese individuals are associated with plasma HDL and glucose levels: a microarray study. BMC Med Genomics. 2010, 3: 34-10.1186/1755-8794-3-34.PubMed CentralView ArticlePubMedGoogle Scholar
- Perez-Enciso M, Ferraz AL, Ojeda A, Lopez-Bejar M: Impact of breed and sex on porcine endocrine transcriptome: a bayesian biometrical analysis. BMC Genomics. 2009, 10: 89-10.1186/1471-2164-10-89.PubMed CentralView ArticlePubMedGoogle Scholar
- Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP: Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 2003, 31 (4): e15-10.1093/nar/gng015.PubMed CentralView ArticlePubMedGoogle Scholar
- R development core team: R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing Vienna, Austria. 2009, ISBN 3-900051-07-0Google Scholar
- Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP: Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA. 2005, 102 (43): 15545-15550. 10.1073/pnas.0506580102.PubMed CentralView ArticlePubMedGoogle Scholar
- Langfelder P, Horvath S: WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008, 9: 559-10.1186/1471-2105-9-559.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhang B, Horvath S: A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005, 4: Article17-PubMedGoogle Scholar
- Tsai S, Cassady JP, Freking BA, Nonneman DJ, Rohrer GA, Piedrahita JA: Annotation of the Affymetrix porcine genome microarray. Anim Genet. 2006, 37 (4): 423-424. 10.1111/j.1365-2052.2006.01460.x.View ArticlePubMedGoogle Scholar
- Huang da W, Sherman BT, Lempicki RA: Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009, 4 (1): 44-57.View ArticlePubMedGoogle Scholar
- Ferraz AL, Ojeda A, Lopez-Bejar M, Fernandes LT, Castello A, Folch JM, Perez-Enciso M: Transcriptome architecture across tissues in the pig. BMC Genomics. 2008, 9: 173-10.1186/1471-2164-9-173.PubMed CentralView ArticlePubMedGoogle Scholar
- Benjamini Y, Hochberg Y: Controlling the false discovery rate: A practical and powerful approach to multiple testing. J Roy Statist Soc, Ser B. 1995, 57 (1): 289-300.Google Scholar
- Yang X, Schadt EE, Wang S, Wang H, Arnold AP, Ingram-Drake L, Drake TA, Lusis AJ: Tissue-specific expression and regulation of sexually dimorphic genes in mice. Genome Res. 2006, 16 (8): 995-1004. 10.1101/gr.5217506.PubMed CentralView ArticlePubMedGoogle Scholar
- Pinto S, Roseberry AG, Liu H, Diano S, Shanabrough M, Cai X, Friedman JM, Horvath TL: Rapid rewiring of arcuate nucleus feeding circuits by leptin. Science. 2004, 304 (5667): 110-115. 10.1126/science.1089459.View ArticlePubMedGoogle Scholar
- Stojilkovic SS: Pituitary cell type-specific electrical activity, calcium signaling and secretion. Biol Res. 2006, 39 (3): 403-423.View ArticlePubMedGoogle Scholar
- Barb CR, Kraeling RR, Rampacek GB: Glucose and free fatty acid modulation of growth hormone and luteinizing hormone secretion by cultured porcine pituitary cells. J Anim Sci. 1995, 73 (5): 1416-1423.PubMedGoogle Scholar
- Barb CR, Kraeling RR, Barrett JB, Rampacek GB, Campbell RM, Mowles TF: Serum glucose and free fatty acids modulate growth hormone and luteinizing hormone secretion in the pig. Proc Soc Exp Biol Med. 1991, 198 (1): 636-642.View ArticlePubMedGoogle Scholar
- Engel HR, Hallman L, Siegel S, Bergenstal DM: Effect of growth hormone on plasma unesterified fatty acid levels of pypophysectomized rats. Proc Soc Exp Biol Med. 1958, 98 (4): 753-755.View ArticlePubMedGoogle Scholar
- Casanueva FF, Villanueva L, Dieguez C, Diaz Y, Cabranes JA, Szoke B, Scanlon MF, Schally AV, Fernandez-Cruz A: Free fatty acids block growth hormone (GH) releasing hormone-stimulated GH secretion in man directly at the pituitary. J Clin Endocrinol Metab. 1987, 65 (4): 634-642. 10.1210/jcem-65-4-634.View ArticlePubMedGoogle Scholar
- Huang MJ, Liaw YF: Clinical associations between thyroid and liver diseases. J Gastroenterol Hepatol. 1995, 10 (3): 344-350. 10.1111/j.1440-1746.1995.tb01106.x.View ArticlePubMedGoogle Scholar
- Moorthy K, Sharma D, Basir SF, Baquer NZ: Administration of estradiol and progesterone modulate the activities of antioxidant enzyme and aminotransferases in naturally menopausal rats. Exp Gerontol. 2005, 40 (4): 295-302. 10.1016/j.exger.2005.01.004.View ArticlePubMedGoogle Scholar
- Li Q, Lee JA, Black DL: Neuronal regulation of alternative pre-mRNA splicing. Nat Rev Neurosci. 2007, 8 (11): 819-831. 10.1038/nrn2237.View ArticlePubMedGoogle Scholar
- Pedrelli M, Pramfalk C, Parini P: Thyroid hormones and thyroid hormone receptors: effects of thyromimetics on reverse cholesterol transport. World J Gastroenterol. 2010, 16 (47): 5958-5964.PubMed CentralPubMedGoogle Scholar
- Lam TK, Schwartz GJ, Rossetti L: Hypothalamic sensing of fatty acids. Nat Neurosci. 2005, 8 (5): 579-584. 10.1038/nn1456.View ArticlePubMedGoogle Scholar
- Williams G, Bing C, Cai XJ, Harrold JA, King PJ, Liu XH: The hypothalamus and the control of energy homeostasis: different circuits, different purposes. Physiol Behav. 2001, 74 (4-5): 683-701. 10.1016/S0031-9384(01)00612-6.View ArticlePubMedGoogle Scholar
- Cheung ZH, Fu AK, Ip NY: Synaptic roles of Cdk5: implications in higher cognitive functions and neurodegenerative diseases. Neuron. 2006, 50 (1): 13-18. 10.1016/j.neuron.2006.02.024.View ArticlePubMedGoogle Scholar
- Wanschers BF, van de Vorstenbosch R, Schlager MA, Splinter D, Akhmanova A, Hoogenraad CC, Wieringa B, Fransen JA: A role for the Rab6B Bicaudal-D1 interaction in retrograde transport in neuronal cells. Exp Cell Res. 2007, 313 (16): 3408-3420. 10.1016/j.yexcr.2007.05.032.View ArticlePubMedGoogle Scholar
- Kitajka K, Sinclair AJ, Weisinger RS, Weisinger HS, Mathai M, Jayasooriya AP, Halver JE, Puskas LG: Effects of dietary omega-3 polyunsaturated fatty acids on brain gene expression. Proc Natl Acad Sci USA. 2004, 101 (30): 10931-10936. 10.1073/pnas.0402342101.PubMed CentralView ArticlePubMedGoogle Scholar
- Yeh TY, Beiswenger KK, Li P, Bolin KE, Lee RM, Tsao TS, Murphy AN, Hevener AL, Chi NW: Hypermetabolism, hyperphagia, and reduced adiposity in tankyrase-deficient mice. Diabetes. 2009, 58 (11): 2476-2485. 10.2337/db08-1781.PubMed CentralView ArticlePubMedGoogle Scholar
- Church C, Moir L, McMurray F, Girard C, Banks GT, Teboul L, Wells S, Bruning JC, Nolan PM, Ashcroft FM, Cox RD: Overexpression of Fto leads to increased food intake and results in obesity. Nat Genet. 2010, 42 (12): 1086-1092. 10.1038/ng.713.PubMed CentralView ArticlePubMedGoogle Scholar
- Dina C, Meyre D, Gallina S, Durand E, Korner A, Jacobson P, Carlsson LM, Kiess W, Vatin V, Lecoeur C, Delplanque J, Vaillant E, Pattou F, Ruiz J, Weill J, Levy-Marchal C, Horber F, Potoczna N, Hercberg S, Le Stunff C, Bougnères P, Kovacs P, Marre M, Balkau B, Cauchi S, Chèvre JC, Froguel P: Variation in FTO contributes to childhood obesity and severe adult obesity. Nat Genet. 2007, 39 (6): 724-726. 10.1038/ng2048.View ArticlePubMedGoogle Scholar
- Cussac D, Leblanc P, L'Heritier A, Bertoglio J, Lang P, Kordon C, Enjalbert A, Saltarelli D: Rho proteins are localized with different membrane compartments involved in vesicular trafficking in anterior pituitary cells. Mol Cell Endocrinol. 1996, 119 (2): 195-206. 10.1016/0303-7207(96)03814-2.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.