A critique of widely used normalization software tools and an alternative method to identify reliable reference genes in red clover (Trifolium pratense L.)

Determination of appropriate reference genes is crucial to normalization of gene expression data and prevention of biased results in qRT-PCR studies. This study is the first attempt to systematically compare potential reference genes to detect the most constitutively expressed reference genes for accurate normalization in red clover tissues including leaves, stems and roots. To identify the best-suited reference gene(s) for normalization, several statistical algorithms such as geNorm, BestKeeper and NormFinder have been developed. All these algorithms are based on the key assumption that none of the investigated candidate reference genes show systematic variation in their expression profile across the samples being considered. However, this assumption is likely to be violated in practice. The authors therefore suggest a simple and novel stability index based on the analysis of variance model which is free from the assumption made by the algorithms. We assessed the expression stability of eight candidate reference genes including actin (ACT), glyceraldehyde-3-phosphate-dehydrogenase (GADPH), elongation factor-1alpha (EF-1α), translation initiation factor (EIF-4a), ubiquitin-conjugating enzyme E2 (UBC2), polyubiquitin (UBQ10), sand family protein (SAND) and yellow-leaf-specific protein 8 (YLS8). Our results indicated that UBC2 and UBQ10 ranked as the two most stably expressed genes in leaf tissue. UBC2 and YLS8 were defined as optimal control genes for stem tissue. EIF-4a and UBC2 were found to be the most stable reference gene for root tissue. GAPDH and SAND showed relatively low stability in expression study of red clover. When all tested tissues were considered, we observed that YLS8 and UBC2 showed remarkable stability in their expression level across tissues.