Hi Ivo,

This is a question that comes up often with my customers. The short answer is that some sort of normalization is always necessary. If I had to hierarchically rank the methods of normalization, use of an endogenous control would probably be at the top of the list. Although it can be difficult to identify a gene that remains stable in some experimental systems (across tissues, etc.), it’s often worthwhile to put in the extra work to do it. I see a lot of reference to the “classical” housekeeping genes – and this is something that gets perpetuated in the literature and on forums – but it’s important to remember that these genes are often not appropriate for use as endogenous controls because their levels fluctuate, and we should really get away from using them by default. We offer 96w or 384w plates (human/mouse/rat) onto which putative endogenous control assays (TaqMan) have been spotted. Several of these genes are common endogenous controls genes, several of them are unique ones; all one needs to do is mix sample/master mix/water, load the plate(s), run the qPCR experiment, and analyze the data (all of our Applied Biosystems software can easily do the analysis). My customers love this product because it can reveal a gene they have previously not thought of normalizing against, one that might be better that others they’ve used previously. Ultimately, this will strengthen the data.

http://www.lifetechnologies.com/order/catalog/product/4396840?ICID=search-product

Sometimes, using an endogenous control for normalization is more difficult, and researchers will look at alternative approaches. 2 examples include normalization by cell number prior to extraction, or normalization against an RNA or DNA that was spiked in prior to extraction. Generally these are all secondary to a stable endogenous control, though. Spectrophotometric quantification of cDNA is notoriously inaccurate, owing to unincorporated nucleotides and other impurities in the mixture, so I wouldn’t recommend that (sorry, I haven’t had an opportunity to read the papers you attached).

Good luck with your experiments.

Hi Ivo, it is a really good question. We work in ecotoxicology and because we find out that usually used endogenous controls varied a lot between gender, tissues and even between differently treated animals, we decided to move to cDNA measurements. The thing is, how many endogenous controls should you measure to find out the less variable one? And what is the maximum value of "less" in that case? The "less" still means variable, and that minimun variability could still mask differences. At least in ecotoxicology each experiment contains a variable that could affect the "commonly used"endogenous gene expression, so, every time that a experiment is planned where qPCRs are performed a proper endogenous gene should be checked. And as result the cost of the experiments increase a lot. Thus, overall, in my opinion is a good alternative. At least for us.

Hi, Oihane, thank you for your answer. If I getting right, you don´t measure endogenous genes any more, do you? What is the cDNA quantification method that you use? Also, We consider that evaluate 2 or 3 endogenous genes only for 1 gen of interest, is too much effort, money, energy, time, etc.

Hi again Ivo. As you will know transitions are not as drastic and fast as we would like. And the methodology has implemented quite recently in my lab. As far as I know (because I am now temporary working in other lab), they´ve been comparing both methodologies to be sure that it will work (I know that a master student has been doing it,), and to convince other  "old-fashion" researchers of the same group that it would work. So, we are now in the transition of  it. Thus, I know that they have normalised the data against ribo-green cDNA measurements, but, I guess that they´ve also measured at least 1 endogenous gene (the less variable one measured in other experiments for the same specie) just in case the reviewers ask it (this is only because the old-fashion ones asked about it!). I have linked the question to some colleagues to see if they could help you better than me. Sorry that I can no longer help you.

Hi Ivo, good question. In some of my work I had the problem that all of the classical housekeeping genes (Actin, Gapdh, Rplp0, 36B4, ... ) where affected by the treatment. Of course, without properly checking cDNA concentration this is an assumption based on the believe that using the same RNA amount as input for cDNA will yield the same amount of cDNA per sample. One thing that would speak in favour of this assumption is, that SD between samples of the same time point and treatment was small and that in time-dependent experiments a housekeeping genes changed over time.

Actually, in one paper - when I was comparing gene expression in different fly tissues - I omitted normalisation to a house keeping gene, and explicitly wrote that "Normalization" was based on total RNA levels for RT only. No one complained, as most people are aware that there are no housekeeping genes that work well across multiple tissues.

It will be great to hear more about peoples experience with the cDNA quantification. I think that it´s superior over house-keeping genes in many cases and better accepted than the "same amount of total RNA approach" only. 

Hi,

I have just had the feedback from my colleagues and they used Quant it Oligreen instead of Ribo-green. Sorry about that.

http://www.lifetechnologies.com/order/catalog/product/O7582

I must say, that we always tend to add the same quantity of cDNA in all of the samples we measured, and we showed a huge variability in our resutls. My colleagues, have compared different housekeepings (the most used ones in our area: 18S rRNA, b-acti, EFalpha, among others) during the whole annual cycle of our target organisms, and in two different tissues (adding the same quantity of cDNA in each reaction). The reason for this experiment set up, is because we work in the monitoring of wild communities and we use to compare different seasons. All measured genes varied and they are confident now about the cDNA normalization methodology. So, they will continue usign this methodology.

As Christoph mentioned it would be great to hear more about different experiences, since I guess that is an intriguing issue that involves different research-areas.

Hi Ivo,

This is a question that comes up often with my customers. The short answer is that some sort of normalization is always necessary. If I had to hierarchically rank the methods of normalization, use of an endogenous control would probably be at the top of the list. Although it can be difficult to identify a gene that remains stable in some experimental systems (across tissues, etc.), it’s often worthwhile to put in the extra work to do it. I see a lot of reference to the “classical” housekeeping genes – and this is something that gets perpetuated in the literature and on forums – but it’s important to remember that these genes are often not appropriate for use as endogenous controls because their levels fluctuate, and we should really get away from using them by default. We offer 96w or 384w plates (human/mouse/rat) onto which putative endogenous control assays (TaqMan) have been spotted. Several of these genes are common endogenous controls genes, several of them are unique ones; all one needs to do is mix sample/master mix/water, load the plate(s), run the qPCR experiment, and analyze the data (all of our Applied Biosystems software can easily do the analysis). My customers love this product because it can reveal a gene they have previously not thought of normalizing against, one that might be better that others they’ve used previously. Ultimately, this will strengthen the data.

http://www.lifetechnologies.com/order/catalog/product/4396840?ICID=search-product

Sometimes, using an endogenous control for normalization is more difficult, and researchers will look at alternative approaches. 2 examples include normalization by cell number prior to extraction, or normalization against an RNA or DNA that was spiked in prior to extraction. Generally these are all secondary to a stable endogenous control, though. Spectrophotometric quantification of cDNA is notoriously inaccurate, owing to unincorporated nucleotides and other impurities in the mixture, so I wouldn’t recommend that (sorry, I haven’t had an opportunity to read the papers you attached).

Good luck with your experiments.

I agree with Mikael. On top of that, normalizing against purified cDNA (to get rid of RNA and primers that may substantially interfere with the cDNA concentration measurements) basically normalizes against ribosomal RNA (that makes up the bulk of the RNA; even when using oligodT primers or no primers at all). It has been shown that rRNA is often not a good choice for normalization (see link to paper attached).

http://www.biogazelle.com/knowledge-center/publications#node-778

Thank you very much everybody for yours answer. Is synthetic DNA or RNAs spike in a good method for normalization or must to be used only for  for quality control throughout  RT-qPCR workflow? It is very common to use syntehetic non-human miRNA spike in to normalize plasma circulating miRNA levels, by qPCR. finally question: Is a geometric mean of 3 or more endogenous control mRNAs the best method to do a relative quantification by qPCR?

Hi everyone, Really great helpful question & answers. I just would like to share my very humble experience with u. I have not used any cDNA quantification, but I did do internal control optimization experiment before being able to find one stable with my treatment. I ended up using two internal controls with every run and I found it useful. what I am doing now and hope to hear ur  opinion about it: I am taking the mean of the two internal controls I am using as they have close ct values. which make me more confident about the results :)

Hi everyone, very interesting an d helpful forum. Personally, I would rather make more controls than trying to avoid them (every experiment is as good as its controls;-). In my lab we try to follow the eleven golden rules for qPCR published in 2008 (see link to paper: http://www.plantcell.org/content/20/7/1736.full. Good luck!

Hi, i think that the most suitable method to normalize your q-PCR data is to quantify the RNA isolated from your samples, because you have to make the same quantity of RNA to get the same standardized quantity for the qPCR.

I agree with Marcel about that controls support the results, however, I want to do just necessary controls, because I don´t want to control the control. I disagree with Younes, because I think that quantify cDNA is better that quantify RNA because the first one is the true template for qPCR reaction, and thus, you can avoid differences in retrotranscription  efficiency. But, as said Jo this methodology only could reflect the ribosomal RNA quantity and no mRNA. This is a real big issue.

Hi, I guess it will depend on the type of analysis you want to use? If it is relative quantification you'll always need an endogenous control do apply the deltadeltaCt method. If it is an absolute quantification qPCR, I guess that would be ok (maybe look for someone that has done it)? I would just look for a really good way to check for cDNA concentration, like the picoGreen kit or so, I would not trust that reverse-transcribing the same amount of RNA would result in exactly the same amount of cDNA, I would measure the cDNA itself, and make sure it is only measuring it, and not the RNA leftovers and dNTPs left by the reaction. Good luck!