Evaluation of potential reference genes in real-time RT-PCR studies of Atlantic salmon

Olsvik, Pål A; Lie, Kai K; Jordal, Ann-Elise O; Nilsen, Tom O; Hordvik, Ivar

To be able to evaluate gene stability under stressful conditions, mRNA expression of the selected genes was examined in gills of salmon going through smoltification. Prior to seawater entry, juvenile anadromous salmon undergo a parr-smolt transformation, characterized by behavioral, morphological and physiological changes, known to be challenging for the fish. Physiological alterations include increased seawater tolerance, olfactory sensitivity, metabolic rate, scope for growth and changed hemoglobin and visual pigments [21] . We selected to examine the gills during smoltification, because this tissue plays a major role in ionic and osmotic regulation during adaptation to hyperosmotic seawater. Figure 1 shows the raw Ct values of the studied genes in gills before, during and after smoltification (smoltified in seawater and desmoltified in freshwater). In Figure 2 the same data are presented, but now normalized against an index calculated by geNorm of the three most stable genes (β-actin, EF1A A and EF1A B ). Based on the M values, geNorm ranks the stability of the six genes from 24 fish going through smoltification in the following order: EF1A B >EF1A A >β-actin>S20>18S rRNA>GAPDH ( Figure 3 ). In Figure 1 it can be seen that the 18S gene had the lowest individual raw Ct variation. Most individual raw Ct variation of the studied genes is seen in the presmolt and smolt groups. A characteristic drop in expression can be seen for all genes in the smolt group, compared to the presmolt group. After transfer to seawater, the individual raw Ct variation decreased for all genes. Overall, the raw Ct data suggest that the physiological challenging smoltification process affected the expression of all six genes. When the same data were normalized against an index of the three most stable genes, β-actin, EF1A A and EF1A B , the relative expression levels were altered for all genes. Now the ribosomal 18S gene emerges as the second worse, whereas the two paralog EF1A genes became the most stable. This might have to do with the fact that geNorm will top-rank co-expressed genes [22] , a weakness that has to be considered when evaluating paralog genes likely to be co-regulated. Even though the eEF1A-2 gene has been identified as an important oncogene and has been shown to be differently expressed in human tissues [11] , Hamalainen et al. [23] found the eEF1A-1 gene to be a good reference gene in real-time RT-PCR examinations. A similar finding was reported by Frost and Nilsen [24] in salmon louse, where they showed that the eEF1A and S20 genes were valid candidate references, whereas the 18S rRNA and GAPDH genes were unsuitable. The current findings based on geNorm evaluation question the recommended application of ribosomal genes as references (as suggested for example by Ambion (see reference [15] ), and are in line with earlier warnings against the use of rRNA genes as references [5, 6] . To avoid the normalization of the genes for β-actin, EF1A A and EF1A B against an index partly based on their own expression, the S20 gene was included in the index instead, and the mean normalized expression for these three genes calculated with the new index. The patterns of expression, however, were approximately the same for the three genes as seen in Figure 2 , suggesting that the gene-stability measure M can be used to find the most appropriate reference genes.