I completely agree with the authors, that control genes e.g. for real time-PCR normalization should be choosen with caution.
However I would like to state that there might be not a single gene that is suitable as universal normalzation control. For example, a gene that is quite stable between selected tissues in vivo could be completely altered under stress conditions like hypoxia or oxidant treatments or differently expressed between cell lines - or altered in a different laboratory environment.
Promient examples are the glycolytic enzyme GAPDH or actin; there are a number of conditions kown where those change in expression; but they might represent suitable control genes in other cases.
Therefore, every researcher should test the stability of his control genes before his own first application in his own set-up ; and reviewers might ask for respective experiments more frequently.
