Comparison of predictive methods and biological validation for qPCR reference genes in sunflower leaf senescence transcript analysis.

The selection and validation of reference genes constitute a key point for gene expression analysis based on qPCR, requiring efficient normalization approaches. In this work, the expression profiles of eight genes were evaluated to identify novel reference genes for transcriptional studies associated to the senescence process in sunflower. Three alternative strategies were applied for the evaluation of gene expression stability in leaves of different ages and exposed to different treatments affecting the senescence process: algorithms implemented in geNorm, BestKeeper software, and the fitting of a statistical linear mixed model (LMModel). The results show that geNorm suggested the use of all combined genes, although identifying α-TUB1 as the most stable expressing gene. BestKeeper revealed α-TUB and β-TUB as stable genes, scoring β-TUB as the most stable one. The statistical LMModel identified α-TUB, actin, PEP, and EF-1α as stable genes in this order. The model-based approximation allows not only the estimation of systematic changes in gene expression, but also the identification of sources of random variation through the estimation of variance components, considering the experimental design applied. Validation of α-TUB and EF-1α as reference genes for expression studies of three sunflower senescence associated genes showed that the first one was more stable for the assayed conditions. We conclude that, when biological replicates are available, LMModel allows a more reliable selection under the assayed conditions. This study represents the first analysis of identification and validation of genuine reference genes for use as internal control in qPCR expression studies in sunflower, experimentally validated throughout six different controlled leaf senescence conditions.