RNA-seq analysis of hepatic gene expression of common Pekin, Muscovy, mule and hinny ducks fed ad libitum or overfed
Duck species are known to have different susceptibility to fatty liver production in response to overfeeding. In order to better describe mechanisms involved in the development of hepatic steatosis and differences between species, transcriptome analyses were conducted on RNAs extracted from the livers of Pekin and Muscovy duck species and of their reciprocal hybrids, Mule and Hinny ducks fed ad libitum or overfed to identify differentially expressed genes and associated functions.
After extraction from the liver of ducks from the four genetic types, RNAs were sequenced and sequencing data were analyzed. Hierarchic clustering and principal component analyses of genes expression levels indicated that differences between individuals lie primarily in feeding effect, differences between genetic types being less important. However, Muscovy ducks fed ad libitum and overfed were clustered together. Interestingly, Hinny and Mule hybrid ducks could not be differentiated from each other, according to feeding. Many genes with expression differences between overfed and ad libitum fed ducks were identified in each genetic type. Functional annotation analyses of these differentially expressed genes highlighted some expected functions (carbohydrate and lipid metabolisms) but also some unexpected ones (cell proliferation and immunity).
These analyses evidence differences in response to overfeeding between different genetic types and help to better characterize functions involved in hepatic steatosis in ducks.
KeywordsRNA sequencing Differentially expressed genes Overfeeding Hepatic steatosis Ducks
Database for annotation, visualization and integrated discovery
Differentially expressed genes
Enriched annotation profiles
Gene ontology biological processes
Non-alcoholic fatty liver disease
Next generation sequencing
Principal component analysis
Polymerase chain reaction
Quantitative polymerase chain reaction
Ribonucleic acid integrity number
Ribonucleic acid sequencing
Trimmed mean of M-values method
Very low density lipoproteins
In western human populations non-alcoholic fatty liver disease (NAFLD), i.e. fat accumulation in the liver, represents the most common cause of abnormal liver function . This hepatic steatosis is associated with overeating of energy-rich food (e.g. high sugar or high fat diet). In terms of human health, NAFLD is frequently associated with various forms of metabolic disorders including obesity, insulin resistance and type 2 diabetes. Increased fatty acids (FA) hepatic synthesis or de novo lipogenesis is known as one major cause of hepatic steatosis and is linked to overexpression of lipogenic genes [2, 3, 4, 5]. In addition, hepatic steatosis is also linked to increased FA uptake, decreased FA β-oxidation and/or decreased synthesis or secretion of very low density lipoproteins (VLDLs) [6, 7]. Expression of adipokines was also demonstrated to play a role in the development of hepatic steatosis [8, 9, 10]. From a clinical point of view, NAFLD is the early step of a sequence of syndromes including steatohepatitis (NASH), fibrosis and cirrhosis and is now considered an inducer of many distinct injurious factors . It is also known that patients with hepatic steatosis do not always develop necroinflammation or liver fibrosis  indicating individual variability in liver disease susceptibility due to genetic and/or environmental factors [13, 14]. Altogether, these results indicate that hepatic steatosis is the result of a large number of metabolic processes, gene expressions and factors that still need to be characterized.
Recently, it was suggested that NAFLD and hepatic steatosis in overfed birds are very similar . Indeed, liver steatosis can occur spontaneously in wild waterfowls as a result of energy storage before migration. This ability is exploited since thousand years in domesticated birds to produce “foie gras” by overfeeding. In these birds, hepatic steatosis exists in isolation, a degenerative event rarely occurs, and it is reversible when overfeeding is stopped . Genetic effects are also known to play a role in waterfowl ability to produce fatty liver [16, 17, 18, 19, 20]. Some evidence suggests that the genetic differences observed in liver steatosis ability between different duck breeds or species are in some part the result of differences in de novo lipogenesis , lipoprotein assembly and secretion [16, 20] and extrahepatic uptake of plasma lipids . Until recently, most of these studies focused on the expression of some genes involved in lipid metabolism as tools for duck genome-wide gene expression were unavailable. It was thus demonstrated that genes involved in carbohydrates and lipid metabolisms were regulated by overfeeding in ducks [21, 22].
The recent emergence of next generation sequencing (NGS) techniques allows the analysis of genes expressed from a whole genome by RNA sequencing. The present study was thus conducted to analyze genome-wide gene expression in Muscovy ducks (Cairina moschata) and common Pekin (Anas platyrhynchos) duck species and in their reciprocal inter-specific hybrids, i.e. Mule ducks (the sterile hybrid from a male Muscovy duck and a female common duck) and Hinny ducks (the sterile hybrid from a male common duck and a female Muscovy duck). Analysis of differentially expressed genes in these 4 genetic types and comparison between species were also conducted in order to evidence similarities and differences of responses to overfeeding in these genetic types that are (Muscovy and Mule ducks) or not (Pekin and Hinny ducks) used for “foie gras” production.
Animals and experimental design
Animals and experimental design have been described previously . Briefly and as described in this publication, male ducks corresponding to common Pekin (Anas platyrhynchos) and Muscovy (Cairina moschata) duck species and to their two reciprocal interspecific Mule and Hinny hybrids (24 per genotype) were provided by Grimaud (Roussay, France). They were reared under usual conditions of light and temperature at the Experimental Station for Waterfowl Breeding (INRA Artiguères, France). From hatching to 6 weeks of age, they were fed ad libitum. From 6 to 12 weeks of age, they were fed on a restricted diet at levels appropriate for each genotype (200–250 g per duck at the beginning, increasing to 360–380 g at the end of the period). At 12 weeks of age, ducks were either fed ad libitum with the growing diet (controls) or overfed 14 days with high carbohydrate corn and corn meal. This overfeeding with high carbohydrate diet is referred thereafter as overfeeding. Ducks were slaughtered 14 h after the last meal by electronarcosis, neck sectioning and bleeding. Immediately after bleeding, liver were weighted and sampled, rapidly frozen in liquid nitrogen and stored at − 80 °C. These liver samples were kindly provided by Baéza and Chartrin  and reused in the present study for RNA sequencing and expression analyses.
RNA preparation and sequencing
Total RNA was extracted from 96 liver samples (12 ducks per genetic type and per diet) using NucleoSpin® RNA L (Macherey-Nagel SARL, Hoerdt, France) according to the manufacturer’s instructions. This kit involves guanidinium thiocyanate, silica membrane and on-column RNase-free DNase digestion. RNA concentration was determined using a ND-1000 Spectrophotometer (Thermo Scientific, Illkirch, France). Integrity of RNA was checked with Lab-on-a-Chip Eukaryote Total RNA Nano and Bionalyzer 2100 (Agilent Technologies France, Massy, France). 79 RNA with absorbance ratio λ260nm/λ280nm and λ260nm/λ230nm > 1.8 and RNA integrity number  or RIN > 7.4 were selected (9–10 ducks per genetic type and per diet).
Libraries preparation and sequencing were performed at GeT Plage, the genomics facility of Genotoul (http://get.genotoul.fr/en/). RNA-seq paired-end libraries have been prepared according to Illumina’s protocols using the Illumina TruSeq RNA Sample Prep Kit v2 to analyze mRNA (Illumina, San Diego, CA). Briefly, mRNA were selected using poly-T beads. Then, RNA were fragmented to generate double stranded cDNA and adaptators were ligated to be sequenced. 10 cycles of PCR were applied to amplify libraries. Library quality was assessed using an Agilent Bioanalyser (Agilent Technologies France, Massy, France) and libraries were quantified by qPCR using the Kapa Library Quantification Kit. RNA-seq experiments were performed on an Illumina HiSeq2000 using a paired-end read length of 2 × 100 pb with the Illumina HiSeq2000 SBS v3 sequencing kits. Samples were tagged using hexamer Tag sequences for subsequent identification (Additional file 1). The libraries were sequenced in paired-ends on 14 different lanes, 6 samples per lane (Additional file 1).
Sequence data were submitted to the NCBI sequence read archive (SRA) under the accession number SRP144764.
Raw sequences pre-treatments and analyses were performed on Bioinformatics facility of Genotoul (http://bioinfo.genotoul.fr/). After FastQC quality check (http://www.bioinformatics.babraham .ac.uk/projects/fastqc/), adapter trimming, and merging sample sequences, RNA sequences were aligned with STAR aligner (version 2.3.0e  on the common duck (Anas platyrhynchos) reference genome (BGI_duck_1.0, INSDC Assembly GCA_000355885.1). A similar approach was applied in a study where horse, donkey and reciprocal hybrids RNA reads were mapped against the horse reference genome . Cufflinks and Cuffmerge tools  were then applied on the merged file to construct a new gene model file. Gene raw counts (Additional files 2 and 3) were obtained using featureCounts (version 1.4.5-p1 using -s 2 -O -p -t exon -g gene_id parameters).
Gene expression analyses
Gene expression analyses were performed using R software version 3.2.2 (https://www.R-project.org). Gene expressions were determined using high-quality reads. Expression data (Additional files 2 and 3) were first normalized with the Trimmed Mean of M-values (TMM) method  using the Bioconductor package edgeR . Normalized data were analyzed by Principal component analyses and hierarchical clustering. Genes were filtered using the data driven HTSfilter procedure using the Bioconductor package HTSFilter . Differentially expressed genes (DEG) between overfed and ad libitum fed ducks were determined in each genetic type using filtered genes set.
Functional annotation of differentially expressed genes with assigned GO terms for biological processes (GOBP) was conducted with the Database for Annotation, Visualization and Integrated Discovery (DAVID, http://david.abcc.ncifcrf.gov/) bioinformatics resource, version 6.82016 [30, 31, 32]. Enrichment of functional annotations in DEG was determined using the default medium classification stringency with an EASE enrichment threshold (a modified Fisher Exact P-Value) set to 0.001. Enriched annotations were then clustered with the Wang method  using GO terms semantic similarity as metric. Expression profiles of down- and up-regulated DEG in the different clusters were also analyzed and visualized using clusterProfiler . Dot size in profiles corresponds to FC. P-values are shown in color bar, values decrease from more (red) to less significant (blue).
Gene interaction networks
Gene interaction networks were determined using the STRING database of known and predicted protein-protein interactions (https://string-db.org/) version 10.5 [35, 36]. Minimum required interaction score was set to medium confidence (0.400).
As previously described , overfeeding induced a significant increase in liver weight, especially in Mule, Hinny and Muscovy ducks and less in Pekin ducks when compared to fed ad libitum control ducks (Additional file 4). This liver weight increase was correlated to increased lipid and triglyceride levels and decreased water and protein levels in the liver. These data indicated that hepatic steatosis occurred more or less in overfed ducks from the 4 genetic types.
RNA sequencing and expression data
Comparison of duck liver transcriptomes
Differential gene expression analyses
Differentially expressed genes
Functional annotation of differentially expressed genes
Some studies have been conducted in duck species and more generally in waterfowls in order to describe hepatic steatosis development after overfeeding. Due to the lack of specific microarrays, most of them were conducted by analyzing the expression of few candidate genes, especially those known to play a role in lipid and carbohydrate metabolisms [21, 22, 38]. The recent development of next generation sequencing (NGS) techniques allows analysis of all the genes expressed from a genome. In this context, we have conducted RNA sequencing and analysis of liver transcriptomes from different genetic types of ducks fed ad libitum or overfed, i.e. Muscovy (Cairina moschata) and Pekin (Anas platyrhynchos) duck species and Mule and Hinny reciprocal duck hybrids.
It appears that gene expression data in these ducks allow to cluster them in 6 distinct groups, according first to overfeeding and then to genetic type, suggesting that differences in gene expression are more due to feeding effect than genetic type effect. Surprisingly, Mule and Hinny hybrids were clustered together in two different clusters according to feeding (Fig. 2). One could expect great differences in gene expression between these two hybrids as fatty liver production is conducted in Mule ducks and not in Hinny ducks. Our results suggest that not using Hinny ducks is not due to their lower ability to produce fatty liver. In fact the reason is more related to the difficulty to produce Hinny ducks according to a less efficient reproduction in terms of fertility and hatchability when female Muscovy ducks are matted to male common ducks . Our results also show that the overfed Pekin group is widely dispersed on dimension 1 with negative and positive values indicating that Pekin ducks have a lower ability and a greater variability to produce fatty liver. The explanation of this greater variability is not clear but could be the result of selection schemes for fatty liver production, mostly conducted in Muscovy parental lines than in Pekin lines.
Differences in gene expressions between fed ad libitum and overfed ducks were then analyzed. To avoid eventual biases due to heterologous mapping and counting of Muscovy, mule and hinny ducks reads on common duck reference genome, DEG were not determined between all genetic types. They were first analyzed in each genetic type. In such a situation biases if any would be identical between ad libitum and overfed ducks from the same genetic type. Then, these DEG were compared between different genetic types (Fig. 3). It appears that many genes were similarly down- and up-regulated by overfeeding in the four genetic types, indicating that response to overfeeding and hepatic steatosis development involve in some part the same genes in the four genetic types. However, some DEG were also found that distinguish ducks according to their genetic types and responses. When enriched functions associated to DEG were analyzed and drawn as enriched annotation profiles the similarities and differences between the 8 conditions were made more obvious showing common and specific enriched functions according to genetic types (Fig. 5). The clustering of these annotations by semantic similarity made it possible to resume the functions associated to the 611 biological processes GO terms in a more synthetic way, i.e. 9 clusters. As expected, some annotations were related to lipid and carbohydrate metabolism (cluster 2) thus confirming and enlarging previous results obtained in ducks [20, 21, 22, 38] and geese [40, 41]. It is interesting to note that together with up-regulation of genes involved in carbohydrate metabolism and lipid synthesis, other genes involved in lipid catabolism were down-regulated, especially in Muscovy, Mule and Hinny ducks (Figs. 7 and 8). These results indicate that hepatic steatosis in ducks involves a lot of genes playing a role in carbohydrate and lipid metabolisms, not only increasing glycolysis and de novo lipogenesis, decreasing lipoprotein assembly and secretion and extrahepatic uptake of plasma lipids as previously documented [16, 19, 20, 21, 22], but also decreasing lipid catabolism. It is also interesting to note that together with up-regulation of lipogenesis and down regulation of lipid secretion and extrahepatic uptake, down regulation of lipid catabolism is more obvious in ducks with a higher ability to produce fatty livers, i.e. Muscovy, Mule and Hinny ducks, and less in Pekin ducks, suggesting that inflammation could be lower in these overfed ducks when compared to Pekin overfed ducks. Muscovy, Mule and Hinny ducks thus appear to store more lipids in the liver than Pekin ducks. Conversely, Pekin ducks export more lipids in the blood, uptake and store them in peripheral adipose tissues . Thought different sensibility to insulin could play a role in the differences observed between genetic types with resistance in Pekin ducks, it appeared however that all genetic types were sensitive to insulin .
Some other unexpected or less documented functions were also found, including down-regulation of inflammation, stress and immune responses, especially in Pekin ducks (cluster 6.1, Fig. 9), and up-regulation of proliferation (clusters 8 and 9). As shown in Fig. 10, down-regulation of genes involved in immune response and inflammation confirms the results of other studies conducted in waterfowls [41, 44, 45, 46]. Knowing the evolution of NAFLD towards more serious liver diseases, we would expect these functions being increased. In the same cluster, down regulation of response to nutrient, exogenous stimuli and insulin were also observed. Resistance to insulin in obese patient is well documented and is linked to type 2-diabetes, hepatic steatosis and other components of metabolic syndrome. We can thus speculate that immune response and inflammation were up-regulated during the first days of overfeeding and were then down-regulated, hepatocytes becoming resistant or insensitive after over-stimulation as observed for insulin response. Analyses of the kinetics of these genes across the entire period of over-feeding would help to address this question. These down-regulations could also be the result of a trade-off between transcription of genes involved in carbohydrates and lipid metabolisms which are dramatically up-regulated and other hepato-specific functions, i.e. expression of complement proteins, which are consequently down-regulated. This hypothesis is reinforced by the fact that albumin, one hepato-specific gene, is also down-regulated after overfeeding ( and this study).
Many GO terms associated to proliferation were also up-regulated (additional file, clusters 9.1 and 2). Our results suggest that in addition to hypertrophy hepatic steatosis is also associated to hyperplasia of hepatic cells. Adipocyte hyperplasia is well documented in obesity [47, 48] but as far we know this has not been reported yet in hepatic steatosis. However, further investigations are now needed to confirm hyperplasia in hepatic steatosis. Other studies should also be conducted to complement our results, focusing on non-coding RNAs knowing their roles in the regulation of gene expression.
Our study is the first report describing whole transcriptomes in four duck genetic types fed ad libitum and overfed. It helps better characterizing responses to overfeeding in ducks in terms of gene expression and associated functions. It also highlights up- and down-regulation of some unexpected functions in duck hepatic steatosis including immunity and cell proliferation. According to duck selection, the results could represent first milestones for early selection in mule’s parents, of ducks with traits of interest to the profession by identifying genes with different levels of expression between genetic types and correlated with their “foie gras” production.
The authors greatly acknowledge Cécile Duby for technical assistance in RNA preparation. The authors also thank the staff of the INRA experimental Palmipèdes à Foie Gras (Palmipôle) unit (UE 0089, Benquet, France) for producing and rearing animals, the staff of GeT core facility (US 1426, GenoToul, Castanet-Tolosan, France) for advices and sequencing, the Bioinformatics core facilities from Genotoul (Toulouse, France) and GenOuest from Biogenouest (Rennes, France) for data storage and access to clusters for analyses.
The research was supported by the National Institute for Agronomical Researches (INRA) AIP Bio-Ressources research grant. With the exception of funding, INRA played no role in this study.
Availability of data and materials
The dataset supporting the conclusions of this article is available in the NCBI Sequence Read Archive (SRA), under the accession number SRP144764 (https://www.ncbi.nlm.nih.gov/sra/SRP144764).
EB: conceived the experimental design and collected liver samples. CD conceived and coordinated the RNA-Seq study. OB and DE generated RNA libraries, sequencing and quality controls of RNA sequences produced. CK designed and implemented the pipeline for RNA-Seq preprocessing, sequence assembly and gene expression counting. FH, MHB and CD carried out RNA-Seq, DEG and annotation analyses. FH, MHB, CK and CD interpreted the data. CD and FH drafted the manuscript. CK, OB, DE and MHB were also involved in drafting some parts of the manuscript and revisited it. All authors read and approved the final manuscript.
Ethics approval and consent to participate
All experimental procedures were performed in accordance with the French National Guidelines for the care and use of animals for research purposes (http://ethique.ipbs.fr/charteexpeanimale.pdf). Ethics Committee authorization to experiment on living animals was not sought for this study as samples were previously collected by another study.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 32.Huang DW, Sherman BT, Tan Q, Kir J, Liu D, Bryant D, Guo Y, Stephens R, Baseler MW, Lane HC, et al. DAVID bioinformatics resources: expanded annotation database and novel algorithms to better extract biology from large gene lists. Nucleic Acids Res. 2007;35(Web Server issue):W169–75.CrossRefGoogle Scholar
- 39.Brun JM, Richard MM, Marie-Etancelin C, Rouvier R, Larzul C. Le canard mulard: déterminisme génétique d’un hybride intergénérique. INRA Productions Animales. 2005;18:295–308.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.