Reverse transcription PCR (RT-PCR) has been used to quantify mRNA (messenger RNA) in gene expression analysis for many years. Researchers generally use the technique to compare two or more samples, for example looking at cancers at different stages, or comparing normal cells and diseased cells, but to control for error in the final results, the samples will need to be normalised.




To make sure that the results are as relevant and useful as possible, it is important to reduce the variability, for example, by handling and preparing the samples in exactly the same way each time, and making sure that they are similar in size (weight or volume of tissue, or number of cells) and quality. Variability can still creep in, through errors in pipetting, the presence of inhibitors, or differences in the RNA extraction, PCR and reverse transcriptase efficiencies.

There are a number of approaches to sample normalisation:

  • Normalise to total RNA
  • Normalise to genomic DNA
  • Normalise to reference gene – rRNA (ribosomal RNA) or mRNA
  • Normalise to an ‘added’ molecule

Normalising to total RNA

RT-PCR uses reverse transcriptase to create cDNA (complementary DNA) from RNA before the PCR step. Measuring the quantity of RNA before reverse transcription ensures that samples start from the same baseline, and it can also confirm the quality of the RNA. However, it doesn’t necessarily negate the effect of any issues in the cDNA preparation or the PCR reactions, and it assumes that there aren’t any variations in the ratio between rRNA and mRNA.

Normalising to genomic DNA

Genomic DNA can be measured directly using qPCR (quantitative PCR). However, genomic DNA can vary in copy number per cell, for example in proliferating cells or in tumour cells, and the processes used to extract the RNA may result in low levels of DNA.

Normalising to reference gene

Normalising to a reference gene or housekeeping gene controls for all the steps in the RT-PCR process, as the reference gene and gene of interest go through the same steps. Reference genes used include:

  • ?-actin
  • glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
  • hypoxanthine-guanine phosphoribosyl transferase (HPRT)
  • 18S ribosomal RNA.

All of these genes are expressed at relatively high levels in all cells, but the reference gene needs to be selected carefully. It needs to be validated and expressed stably in the target cells, and any variability in expression must be taken into account. Using multiple reference genes can improve accuracy, but mean that assays will take longer and cost more.

Normalising to an ‘added’ molecule

Adding in a synthetic RNA, or one cloned from another organism at the beginning of the procedure will control for all the steps in the RT-PCR process, without the variability of an endogenous RNA. This RNA will still need to be validated, however, as it is not be extracted from cells like the test RNA.

All of these approaches have advantages and disadvantages, and need to be considered on a case-by-case basis, with appropriate validation to ensure that the resulting data is high quality, relevant and useful.

Further reading: Real-time RT-PCR normalisation; strategies and considerations

Suzanne Elvidge is a freelance science, biopharma, business and health writer with more than 20 years of experience. She is editor of Genome Engineering, a blog that monitors the latest developments in genome engineering and that aims to educate (and sometimes to entertain!) and has written for a range of online and print publications including FierceBiomarkers, FierceDrugDelivery, European Life Science, the Journal of Life Sciences (now the Burrill Report), In Vivo, Life Science Leader, Nature Biotechnology, PR Week and Start-Up. She specialises in writing on pharmaceuticals, biotechnology, healthcare, science, lifestyle and green living, but can write on any topic given enough tea and chocolate biscuits. She lives just beyond the neck end of nowhere in the Peak District with her second-hand bookseller husband and two second-hand cats.