Dear Cristina,

in my experience, the best way to address this issue is testing a few reference genes and calculate the stabilty of these among your samples. To choose the best candidates to normalize your samples you could use a software.

I use qBase, that scores the references genes, and it will help you to select the right ones in your experiment, to compare samples from different tissues and/or treatments. There is a 15 days free trial and the support of the Biogazelle company is quite useful: http://www.biogazelle.com/products

I would reconmend you as well this free software (BestKeeper) where you can score the reference genes: http://www.labtools.us/bestkeeper/.

And, about the Ct values in the reference genes and genes of interest (GOI), if the tissue has an effect in the RNA isolation procedure (pcr inhibitors, lower RNA yield...), it would affect both housekeeping and GOI in the same way, then, with the normalization you would correct these differences among tissues.

I hope this helps,

Aida

Dear Cristina,

in my experience, the best way to address this issue is testing a few reference genes and calculate the stabilty of these among your samples. To choose the best candidates to normalize your samples you could use a software.

I use qBase, that scores the references genes, and it will help you to select the right ones in your experiment, to compare samples from different tissues and/or treatments. There is a 15 days free trial and the support of the Biogazelle company is quite useful: http://www.biogazelle.com/products

I would reconmend you as well this free software (BestKeeper) where you can score the reference genes: http://www.labtools.us/bestkeeper/.

And, about the Ct values in the reference genes and genes of interest (GOI), if the tissue has an effect in the RNA isolation procedure (pcr inhibitors, lower RNA yield...), it would affect both housekeeping and GOI in the same way, then, with the normalization you would correct these differences among tissues.

I hope this helps,

Aida

Hello. Are you able to use a few housekeeping genes such as B-actin and GAPDH? They should be expressed in both liver and adipose and ought to give an indication of the extraction efficiencies from these samples.

I would also recommend you to make your experiment in two times.

- First select several housekeeping genes, I would say 10. Test them in all samples. Use qBase or BestKeeper (thanks to Aida) to select the best 3 stable genes.

- Finally, do your experiment on your candidate genes and include the 3 best housekeeping genes in your study for further normalisations.

This way will makes your experiment really strong in case of paper submission.

Thank you Aida and Caroline,

I have tested GAPDH, TBP, B2M and ACTB in different experiments. I have evaluated their stability with genorm software, and all of them are good if I look the stability M measure of this software. But these softwares do not take into account the treatment/tissue, and when I analyse for differential expression between tissues I find a significant effect of the the tissue on the expression of most or all of them (depending on the experiment). Usually one would not use a reference gene affected by the treatment, but in the case of the tissue I doubt if (assuming that all the tested genes are stable according to usual softwares) the good endogenous gene is the one which is not DE by tissue or the one that actually is DE....

In other words: If I find that GAPDH has lower Ct values in liver I do not know if this is due to higher expression in liver (then it will not be a good reference gene) or higher quality of liver cDNA (and then it will be a good reference one). I think that it is more than probable that liver cDNA is better than adipose cDNA, then I would expect that a good reference gene, whose expression is constant across tissues, would have lower Ct values in liver... Do you agree with this?....

At the UCSC genome browser website, you can find results of microarray expressions for genes in many tissue.

A quick look at it, will gives you the answer that GAPDH is not equally expressed in Liver vs Adipocyte.

Based on this website, you will be able to get a first screen of potential stable Housekeeping genes.

Then you will be able to test these genes in your candidate tissues by qPCR.

Here is the example of the GAPDH gene on UCSC genome browser.

Just take a look at the Microarray Expression Data section.

http://genome.ucsc.edu/cgi-bin/hgGene?hgsid=327223461&hgg_section_ctd_close=1#ctd

Hi Cristina! I agree with you about the softwares... We test ten reference genes with tree different samples for every experiment/treatments/tissues. Then we analyse the raw data and choose the one with Ct curves in same range (or order of magnitude), . I discard any outlier. If I have some interference from pippeting, quantification or something else, it will be around the same for everyone in the experiment, so I discard the normalization for testing endogenous control but always observe the Ct values in all the tests to check that they do not change.

Hi Christina,

I also generally use qBase to find the most suitable ECs for a certai nexperimental setup. However, this strategy doesn't work for all types of experiments; i.e. gene expression profiles of iPSCs, ESCs, or adult stem cells that undergo differentiation. Another option for normalization is "global mean" in case you have a high number of genes you are probing or another alternative is normalization against expressed alu repeats. See the references below.

This link might also be of interest for you:

http://www.biogazelle.com/presentations/20120704_Heidelberg_normalization.pdf

D'haene B, Mestdagh P, Hellemans J, Vandesompele J. miRNA expression

profiling: from reference genes to global mean normalization. Methods Mol Biol.

2012;822:261-72. doi: 10.1007/978-1-61779-427-8_18. PubMed PMID: 22144205.

Willems E, Leyns L, Vandesompele J. Standardization of real-time PCR gene

expression data from independent biological replicates. Anal Biochem. 2008 Aug

1;379(1):127-9. doi: 10.1016/j.ab.2008.04.036. Epub 2008 Apr 26. PubMed PMID:

18485881.

Marullo M, Zuccato C, Mariotti C, Lahiri N, Tabrizi SJ, Di Donato S, Cattaneo

E. Expressed Alu repeats as a novel, reliable tool for normalization of real-time

quantitative RT-PCR data. Genome Biol. 2010 Jan 28;11(1):R9. doi:

10.1186/gb-2010-11-1-r9. PubMed PMID: 20109193; PubMed Central PMCID: PMC2847721.

Good luck!

Cheers,

Andy

I think GAPDH has been shown to change expression with changes in p53 levels, B2M is part complement so if you are studying any process that involves or might involve complement you may see a change.

BestKeeper seems like a great idea. Have you tried adding the sample data (non-controls) and analyzing it with the controls using the software? Do you have microarray data on these samples. We normally look for genes that change expression but you can also use the data to look for potential genes that do not change expression at all.

Hi Cristina,

I second Christophe Poulet's idea. It is always a good idea to search an expression database before designing the experiment. You might be able to find studies with similar experimental conditions a get a sense of which genes remain unchanged. Please, give it a try to the NCBI's Gene Expression Omnibus database

GEO DataSets http://www.ncbi.nlm.nih.gov/gds/

GEO Profiles http://www.ncbi.nlm.nih.gov/geoprofiles/

Here are the field tags you can use to search the database http://www.ncbi.nlm.nih.gov/geo/info/qqtutorial.html#fields

Good luck!

Consitutive expression of the histone variant H3.3 (hv2) is essential in Tetrahymena (Yu and Gorovsky (1997) Mol. Cell. Biol. 17:6303. We have found this gene to be expressed at about the same rate from starvation through all the steps of mating. Yin et al., Eukaryotic Cell (2010) Eukaryotic Cell 9:1343. You might see whether there is a homologous essential histone variant in your organism.

Cristina,

We are using some selected by a group that published in PLOS One, where they examined the problem directly. Check out this paper

de Jonge HJM, Fehrmann RSN, de Bont ESJM, Hofstra RMW, Gerbens F, et al (2007) Evidence Based Selection of Housekeeping Genes. PLoS

ONE 2(9): e898. doi:10.1371/journal.pone.0000898

Michael's Bradbury's comment support our (Poulets and miy) previous comment, they do a database search and find those that remain unchanged. Here the quote:

"Microarray expression data of 13,629 publicly available samples hybridized to Affymetrix HG-U133A and HG-U133 Plus 2.0 GeneChips (Affymetrix, Santa Clara, Ca.) were downloaded from the Gene Expression Omnibus.[14] "

Our ten tested genes has been choosen from microarray databasis also.

Hi, you may try NORMA-gene. Heckmann et al. BMC Bioinformatics 2011, 12:250

Dear Cristina,

Indeed, as written Aida Rodríguez, the best way is to use two or more reference genes. I use GeNorm Plus software of the Biogazelle company where you can score the reference genes. More about expression patterns and stability of the reference gene expression in different tissues you can found on the resource www.http://biogps.org.

Tim

I agree with Timur.It is preferable to use at least two reference genes upon checking of RIN value after extraction of RNA for different tissues.

Dear Chrsitina

I agree with the other; You need to try several genes and then find out the best ones, for exemple following "normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes.

Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F.

Genome Biol. 2002 Jun 18;3(7):RESEARCH0034. Epub 2002 Jun 18."

or any other programm you want to.

I agree with these comments. In fact, you have to try several House Keeping Gene (HKG) in your different conditions and select at least 2 HKG. And this, each time than you want to analyse a new gene of interest in a new condition. But be careful when you choose the softwares. They give you different values to estimate reference gene expression stability. For geNorm, you have Mean M (average expression stability). For BestKeeper, this is r (all candidates reference genes are correlated). For NormFinder is the stability value and for hkgFinder the fold change. You see, you have the choice!

since genes of interest may have varying levels of expression. I find that choosing the normalizing genes accordingly is a better idea. in my experience a combindation GAPDH, hgprt , some ribosomal proteins will be better

In our experience one of the housekeeping genes like HSP70 is a valid control for gene expression normalization. Important is to quantify carefully the expression signal with either laser-scanning densitometry or Q-PCR.

Buona fortuna.

Cristina, we follow the geometric average of 4 different house keeping genes. here is the link

http://genomebiology.com/2002/3/7/research/0034

Thanks a lot for all your comments!!

While it seems intuitive to use a few genes to normalize across samples in gene expression studies, you should keep in mind that every gene expression value made is made with measurement error and/or sample-specific/chip-specific bias. This includes the few you assume are a valid normalization set and so-called 'housekeeping' genes. When you then use this smaller set, the random error and bias from the normalization set is then applied either to all the gene expression measurement values taken. This adds noise, for certain and while it may be hoped that a given normalization method might remove bias, there are instances when the noise added is large and become a source of bias itself. Thus such normalizations are 'lossy transforms' in information theory. If the species you are comparing are closely related, global normalization methods may be attempted. Please see Jordan et al. (2008) for a method to evaluate the effects of normalization and transformation methods on your data. While designed for microarray studies, if your panel is large enough, you might find simpler methods to cause less information loss. Good luck!

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2623303/