Achieving reliable results using qPCR is only possible after application of appropriatenormalisation methods. This is a crucial requirement since qPCR technique impliesproblems at different steps: Ribonucleic acid (RNA) extraction procedure, samplestorage, sample quality, reverse transcription with complementary DNA (cDNA)synthesis including non-optimal primers and inappropriate statistical analysis (Bustin etal., 2009; Mallona et al.
, 2010). Even at the stage of obtaining material from tissuesthere could be a problem in getting samples containing the same amount and quality ofmatrix for the reaction (Huggett et al., 2005). Errors in pipetting and transferring of thematerial may be one of the reasons for variability among samples. The process ofextraction and purification of RNA may proceed with varying efficiency — it is relatedto instability of NA. In addition, biological material often contains many low-specificNA that are a part of the background not to mention native RT and PCR polymeraseinhibitors (Bustin, 2004). Considering all the problems the normalisation step isessential. Most authors agree that the use of reference genes is the most effective methodand is likely to be one of the easiest one to correct errors for valid research results(Huggett et al.
, 2005).Reference genes are an internal reaction control that have sequences different than thetarget. A gene that can be used as appropriate reference must meet important conditions(Chervoneva et al.
, 2010). The most important is that its expression level cannot beaffected by experimental factors. Furthermore, the expression level of that gene shouldbe more or less stable between tissues and physiological states of the organism. It isdesirable to pick such reference that would show a similar threshold cycle with the geneof interest.
It seems that the perfect fulfilment of these conditions are found in the basicmetabolism genes (called Housekeeping Genes – HKGs) which, by definition, areinvolved in processes essential for the survival of cells and must therefore be expressedat a non-regulated constant level. In fact, they were the first to be examined as referencegenes (Thellin et al., 1999).Target genes and reference genes have two major sources of variability, experimentalvariability associated with the technology and the innate or natural variability of thegene (between tissues, individuals, treatment etc.).
The original approach fornormalisation was to find a single reference gene with the most stable expression (in thesense of the smallest variability) across tissues and individuals (Chervoneva et al.,2010). Starting with the work of Vandesompele and colleagues (2002), normalisation iscarried out using a geometric mean (inverse natural logarithm of the mean of the log-transformed gene expressions) of multiple internal control genes as a normalizing factor.The rationale is that the same experimental error should be present in all genesexpressed in the same sample, if all genes are processed simultaneously (Vandesompeleet al., 2002). Thus, the experimental errors of individual replicates are averaged acrossthe reference genes, and a geometric mean of reference genes provides a more robustestimate of the experimental error than individual reference genes (Chervoneva et al.
,2010). Most authors agree that use of single reference gene is not enough to ensurerobust quality for estimating gene expression, thus it is recommended to use at least thetwo most stable reference genes validated for a given experiment design. The validity ofa test can also be enhanced choosing reference genes from different functional classes sothat the chance of co-regulation among those genes is reduced (?y?y?ska-Granica andKoziak, 2012).