Experimental design
Ninety rainbow trout (INRA delayed-spawning trout line) were raised in freshwater tanks for several weeks. Breeding and challenge tank contains between 300 to 500 liters (fish density approx. 10 g.L-1). Mean weight of the fish is estimated to 100g. Fish were kept at natural photoperiod (14h30 light and 9h30 dark) during the experiment. The challenge consists in a freshwater to seawater (32 PSU) transfer. At each time point, fish were removed from the tank and anesthetized in less than 1 minute in 10mL / 10L phenoxyethanol solution.
Chapitre II - Analyse transcriptomique différentielle
For each condition were sampled 2 mL blood (with a seringue in the cuverian vein) and 200 mg gills tissue on 6 fish before and 2, 8, 24, 48, 96 (4 days), 288 (11 days) and 552 (23 days) hours post transfer (PT).
Plasma analysis
Blood was centrifuged (10 min., 2500G, 4°C) and isolated plasma kept at -20°C.
Cortisol concentration was determined by radio immunoassay as described by Auperin and Geslin (2008) and chloride using a colorimetric method (chlorides LDM SOBIODA®).
Statistical analyses on plasma results were performed using Statistica 7.0 software (Statsoft, Tulsa, OK). Regarding data, differences between groups were analyzed using non parametric U Mann-Whitney test.
RNA purification, cDNA synthesis
Pieces of gills (near 200 mg), were removed, and total RNA was extracted by the single step guanidium isothiocyanate-phenol-chloroform method (Chomczynski and Sacchi, 1987) using 2mL TRIzol per sample. RNA integrity and concentration were controlled with an Agilent 2100 bioanalyser (Agilent, Palo Alto, CA).
Microarrays processing. Methods for microarray experiments (cDNA microarrays production, cDNA target preparation, hybridization, signal processing, data analysis, identity of micro-array cDNA clones) were carried out as described in detail by Rescan et al. (2007) with a few changes. In brief, nylon micro-arrays (7.6 x 2.6 cm) were obtained from INRA- GADIE resource center (Jouy en Josas, France). This AGENAE nylon membrane contains 9024 distinct rainbow trout cDNA originating from pooled-tissues libraries (Govoroun and Guiguen, 2006) and finally have 9217 spots corresponding to 8345 different annoted contigs.
cDNA were prepared from 5 µg of total RNA of each sample by simultaneous reverse transcription and labelling. The first step of reverse transcription consist in mixing total RNA with 0.3ng of luciferase control RNA and 0.8µg of dT25, followed by a denaturation step at
Chapitre II - Analyse transcriptomique différentielle
70°C and annealing step at 42°C. Labelling was done for 2 hours at 42°C in the presence of 30 µCi [alpha-33P] dCTP, 2,4 µM dCTP, 400 µM each dATP, dTTP, dGTP and 200 units Superscript II reverse transcriptase (Invitrogen). RNA was degraded (20 min., 68°C, 0.3%
SDS, 15mM µl EDTA and 250mM NaOH), and the remaining cDNA equilibrated at room temperature for 10 min. Neutralization was done by adding 200mM Tris and 0.12N HCL. To block polydT sequences, 2 µg of A80 were added (5 min. 99°C then 1 h at 65°C).
A first hybridization was performed using a 33P-labelled oligonucleotide (TAATACGACTCACTATAGGG) which is at the extremity of each PCR product to determine the amount of cDNA present in each spot. After stripping (3 hours 68°C, 0.1 SSC, 0.2% SDS), arrays were prehybridized (4 hours 65°C in 5x Denhardt’s, 5 SSC, 0.5% SDS) and hybridized with denatured labelled cDNAs (48 h 68°C in 5x Denhardt’s, 5 SSC, 0.5%
SDS). After 3 washes (1 h 68°C, 0.1 SSC, 0.2% SDS), arrays were exposed to phosphor- imaging plates before scanning (FUJI BAS 5000).
Quantitative Polymerase Chain Reactions (Qpcr). For Qpcr, 5 μg of total RNA were reverse transcribed using 200 units of MMLV reverse transcriptase (Invitrogen, Cergy Pontoise, France) and 250 ng random hexamers (Promega) in a reverse transcription master mix containing 2 mM dNTPs, 50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl2, 10 mM dithiothreitol, pH 8.3. Twenty-five units of RNase inhibitor (RNasin, Promega) were added to the reaction. RNA and dNTPs were denatured for 5 min at 65°C, and then chilled on ice before addition of reverse transcription master mix. Reverse transcription was performed at 25°C for 10 min then at 42°C for 50 min followed by a 15-min incubation step at 70°C.
Control reactions were run without reverse transcriptase and used as negative controls in the real-time PCR study for all target genes. QPCR was performed on a Step One Plus real-time PCR system (Applied Biosystems, Foster City, USA). Reverse transcription products were diluted to 2%. Triplicates were used for each sample. Real-time PCR was performed using a
Chapitre II - Analyse transcriptomique différentielle
real-time PCR kit provided with a SYBR Green fluorophore (Fast SYBR Green Master Mix, Applied Biosystem). Primer concentrations were 200 nM to 600 nM depending on the target genes (see Table 1 for details). The hot start enzyme was activated 20 seconds at 95°C, then the amplification was carried out using the following cycle: 95°C for 3 sec; 60°C for 30 sec;
40 times. A pool of reverse transcribed RNA was serially diluted and used to calculate a standard curve. For the choice of the housekeeping genes, we first selected severals genes with a minimum expression variation between freshwater and seawater fish in the microarray results and 2 frequently used housekeeping genes (ef1α and 18S). We thus defined 5 potential housekeeping genes (ef1 alpha, 18s, Ribonuclease P protein subunit p21, Serum paraoxonase / arylesterase 2 and proto-oncogene tyrosine-protein kinase Src). These genes were tested under GeNorm (Vandesomple et al., 2002) which revealed ef1 alpha and proto-oncogene tyrosine-protein kinase Src to be the best internal standards to normalize the cDNA amount of candidate genes. After amplification, a melt curve was obtained according to the following protocol: 10 sec holding followed by a 0.05°C increase, repeated 80 times, and starting at 55°C. Primer sequences of studied genes are presented in Table 1. The control reactions were used to calculate background expression level for each gene. Statistical analyses on QPCR results were performed using Statistica 7.0 software (Statsoft, Tulsa, OK). Regarding data, differences between groups were analyzed using non parametric U Mann-Whitney test.
Microarray analyses. Transcriptomic data normalization consists on a centering by membranes creation groups and by sample add to a vector correction. After eliminating data with wrong signal, a final data set with 74 membranes and 7280 clones were kept for statistical analysis (3 to 6 samples by condition and by time). Statistical analysis consists in a time by time and global freshwater versus seawater Significance Analysis of Microarray performed with Sam Microsoft Excel package (SAM, Tusher et al., 2001). To correct for multiple testing and have information concerning false positive findings, we used FDR (False
Chapitre II - Analyse transcriptomique différentielle
Discovery Rate) method (Benjamini et al., 2001). Two levels of different stringency were then applied. A first SAM analysis with a FDR fixed to 0% provides us a core of the most significantly differentially expressed genes, from which relevant discussion about biological processes and functions occurring after a salinity transfer can be drawn. A second less stringent analysis with a FDR fixed to 5% will provide us a larger number of genes potentially involved in the hyperosmotic response. This gene set allows us to support some functional findings of the FDR 0 results as well as to investigate more global functional vision by Gene Ontology enrichment analysis.
Clustering using the Bootstrap method on TMev (Tigr MultiExperiment Viewer 4.1) allows us to reveal the existence of important genes groups, characterized by similar expression profiles. We here use this approach on the gene differentially found expressed by the fdr 5 SAM analyses. A minimum threshold for the bootstrap value on 90 with more than 2 entities by cluster is fixed.
Initial annotation used in this study is based on the rainbow trout SIGENAE s.om.10 assembly (http://www.sigenae.org). This database is based on Uniprot information with best description found in non-mammals and mammals database. Gene ontology was studied by using AMEN software (Chalmel and Priming, 2008) and a more specialized fish annotation described in Rolland et al. (2009). Briefly, genes / transcripts corresponding to the EST sequences, provided by http://www.sigenae.org website, of the 9024 clones spotted on the Agenae microarray were annotated by searching for potential orthologous genes in the genome of sequenced model fish species (Gasterosteus aculeatus, Danio rerio, Oryzias latipes and Takifugu rubripes). Overall, 8197 trout clones were successfully linked to a fish gene with at least 2 EST hits across all 4 genomes leading to the same Ensembl gene or to similar genes encoding proteins belonging to the same family. The 8197 clones finally
Chapitre II - Analyse transcriptomique différentielle
associate clones with GeneOntology (GO) terms (the Gene Ontology project, 2008). GO terms associated with corresponding mammalian genes (rat, mouse and human) were also used. Thus, 7065 clones present associated fish genes with mammalian orthologous genes, as predicted by the Ensembl database (Compara, version 52). This strategy permits to annotate 88% of the spotted clones in terms of "biological process", "molecular function" or "cellular component".