I performed real-time PCR and I was looking for expression fold changes for 2 genes and I had two sample pools, one treated and the other not treated (for each gene). The problem is that my housekeeping gene (UBQ 30) got a fold change on the treatment. Therefore, what is the correct interpretation and data treatment?

I mean, is it correct to use each control and treatment I got for UBQ CT values to normalize the other two genes corresponding CT, as housekeeping genes are "by definition" expected not to change? (and that is why indeed they are used to normalize)

or should I rather make a mean of my whole UBQ CT-s and normalize all other genes CT values with that mean?

  • $\begingroup$ In principle the housekeeping genes should be constant. But: Changes, that happen to the whole cell, can happen. So it is advisible to have more than one housekeeping gene if possible. $\endgroup$
    – Chris
    Commented Mar 17, 2014 at 16:46

3 Answers 3


You should try another housekeeping gene like GAPDH since the very definition of these genes is that they shouldn't change with your treatment

  • $\begingroup$ good advice - this is a list of even better ones...ncbi.nlm.nih.gov/pubmed/17394773 $\endgroup$
    – shigeta
    Commented Mar 19, 2014 at 3:57
  • $\begingroup$ There is quite a list of genes that can be used - sometimes even GAPDH or Actin change. This needs to be verified for each experiment. $\endgroup$
    – Chris
    Commented Mar 19, 2014 at 6:32
  • $\begingroup$ and how could i objetively discrimine if a given change for CT values is significable as to consider to change my housekeeping gene? $\endgroup$
    – Katz
    Commented Mar 19, 2014 at 11:07
  • $\begingroup$ As Chris says you will have to verify that your housekeeping gene doesn't change between experiments. $\endgroup$
    – V_ix
    Commented Mar 19, 2014 at 14:34

As pointed out by others a reference gene should not change its expression between the control and test samples. Reference gene need not be a housekeeping gene; housekeeping genes are generally used because they are present in all cells are are not subject to any specific regulation. However, global changes in the cell, such as stress or development, can affect housekeeping genes.

When you are not sure about which gene to choose as a reference then you can use a spike in control.

The change in your experiment could be a result of human error. You should replicate the experiment to see if the trend remains the same. If your gene shows consistent trend in all replicates then the change is not likely because of human error.

In general you should choose the reference gene based on your prior knowledge about the condition that you are testing. For example if I am studying cell migration then I would not use Actin as a control. If you are studying processes that may affect metabolic rate then you should perhaps not use GAPDH as a control.

Another good reference gene(s) that is also constitutive would be that for a ribosomal protein.

Also have a look at this post for statistical analysis of RTPCR data.


I'm gonna throw a curve ball here. :) I think the "housekeeping" or "reference gene" concept is fundamentally flawed, because it rests on circular logic. In order to verify that a "housekeeping" gene X is constant across your samples, you must assume that the $C_t$ values themselves are actually a valid measure, so that you can assess the expression level of X directly. But if $C_t$ values are assumed valid, then you don't need a housekeeping gene in the first place! This procedure assumes what it intends to prove; therefore it is circular.

The normalization problem is just not solvable without additional assumptions. A spike-in control works if you are worried about technical variation, but it does not address biological variation. A reasonable assumption I think is that the total mRNA pool is constant; this underlies most normalization methods used in microarray or RNA-seq analysis. For RT-PCR, this assumption can be tested using cDNA dyes, as in this paper. (cDNA is advantagous rather than isolated RNA, because the reverse transcription step is notoriously noisy and should be controlled for.)

In my mind, housekeeping genes are only useful it you really know in advance that they must be constant, and do not need to check this assumption; then the logic is not circular. In this case, if you observe changing $C_t$ values for the housekeeping gene, then you must conclude that the experiment went wrong somehow, and the data should be discarded. But I think this is rarely the case in practise.

This is not the most common opinion on this often-debated matter, but I think it is the right one. :)

  • $\begingroup$ 1. Regarding the vicious cycle of reference gene selection: Ideally there should not be a change in Ct value if there is no change in expression. There may be no need to have a reference gene if situation was always ideal. Here, reference gene simply does the job of a loading control. The perfectly done experiment may not need one but getting that perfection is hard; but once you have established a right reference you can forget about perfection. $\endgroup$
    Commented May 28, 2015 at 7:18
  • $\begingroup$ 2. Regarding spike-in: In cases where there is biological variation of the gene pool, spike-in would be particularly useful. For example, during growth/development a lot of proteins accumulate, including housekeeping genes. They may even be subject to regulation. Spike in would let you know the absolute change in the RNA expression. In these cases the housekeeping gene can also be used to understand correlation and differentiate general regulation from specific regulation. In certain cases you can transfect the spike into the cells prior to RNA isolation. $\endgroup$
    Commented May 28, 2015 at 7:27
  • $\begingroup$ I however concur that the choice of reference is dependent on prior knowledge about the condition. +1 for the alternate opinion :) $\endgroup$
    Commented May 28, 2015 at 7:28
  • $\begingroup$ @WYSIWYG, regarding (1), my point is that in order for a gene X to be a control for suspected variation in Ct, we must assume that gene X is constant, and we cannot verify this by measuring Ct; that becomes circular. For (2), yes spike-ins allows quantitation, but doesn't solve the normalization issue, you still need to relate the measurement to something --- tissue weight, cell number, total protein, or RNA (which I suggested). $\endgroup$
    – Roland
    Commented May 30, 2015 at 20:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .