Aref,  There is no such thing as the best housekeeping gene. The ridiculous numbers of papers studying this is proof that no one is fully satisfied.   All gene expression changes with the proliferation and differentiation of cells.  Housekeeping genes are defined as those that change the least, but which one will be best depends on the cells you use as indicated by Adam.  I think it would be good to use several housekeeping genes to help you see what might be a significant change in your gene versus modest changes in  housekeeping genes. 

What type of cells are you using?

Aref,  There is no such thing as the best housekeeping gene. The ridiculous numbers of papers studying this is proof that no one is fully satisfied.   All gene expression changes with the proliferation and differentiation of cells.  Housekeeping genes are defined as those that change the least, but which one will be best depends on the cells you use as indicated by Adam.  I think it would be good to use several housekeeping genes to help you see what might be a significant change in your gene versus modest changes in  housekeeping genes. 

Dear Aref

there is much talk about which housekeepers to use - to that end see literature I have provided in the link to my drop box folder on qPCR housekeeping genes -  but in my experience GapDH tends to be my favoured choice (because it generally works consistently and ubiquitously well) but I have also elected to use b tubulin and b actin as well

There is an accepted concensus that use of 3 housekeeping genes and computation of the resulting geometric mean provides the most stable and accurate reference gene data; See for example the program 'Gene Norm' and associated literature:

http://medgen.ugent.be/~jvdesomp/genorm/

To take a more empirical approach however you could and indeed I have done the following:

1. Look at precedents for your cell types in the literature as implied by Adam above

2. Try performing qPCR with one or two housekeeping genes -preferably selected from the literature and optimal for your tissue/organism types -  on the same plate and then carry gene expression analysis. You could wel, find that the relative gene expression you derive from the 2 housekeppers iis identical in terms of trends in gene expression: Personally I have done this a number of times comparing data from GapDH and B tubulin and have found identical expression trends; Indeed in most cases identical expression magnitudes from the 2 independent sets of housekeepers

3. My own feeling rooted in emprical practice is that as long as you can demonstrate that the ct values you derive from 1 or more housekeeper(s) are identical within and between sets of biological replicates, treated and untreated and that this remains true for independent assays, i.e. that your intra and intersample variability (repectively) is minimal then I would conclude that your housekeeper is valid for your own perrsonal gene expression studies. I have generated much qPCR data using this maxim and indeed have used just one housepper to derive actual data validated this way

I would say how ever that literature precedents is a good place to start to try and pick an optimal housekeeping gene, and if an identical tissue + organism is not reported in published paper, then at the very least there is plenty of literature regarding optimal housekeepers for particular tissue types and, in fact studies, of this type are provided in one of the links I have provided. Failing that simply try GapDH and b Tubulin and  validate  as described

https://www.researchgate.net/post/How_do_I_include_the_standard_deviation_of_several_Housekeeping_genes_in_the_relative_expression_levels_of_the_gene_of_interest_in_qPCR_analysis?_tpcectx=profile_answers

https://www.dropbox.com/sh/o2qafw0ag6klwd9/AACuGRKo_jPugZWkJ7F7lP1ya?dl=0

Hi Aref, 

I completely agree with comments above that the best housekeeping gene  is the one that changes the least in your particular cell type and treatment/condition you are testing. 

In my experience (T cells cultured in vitro), among HPRT, GAPDH, b-actin, PoI II and TBP, TBP was most reliable, although when possible I try to used two, b-actin and TBP. 

Best

The name "housekeeping gene" in the context of real-time PCR is already misleading.

This term came up with the advent of PCR to signify genes that were expressed ubiquitously, in any cell type in any developmental stage at any time. Just to have a common positive control for the PCR (it was hell expensive to make new and different primers for this purpose every time!).

It was never expressed that the expression of such "housekeeping genes" should be constant, similar, or comparable between different celltypes, depelopmental stages, times, or treatments!

Real-time PCR requires a loading control. This is something entierly different. I have no idea why the fist users went for using housekeeping genes for this purpose. It is surely a neccesity that the gene used as loading control is expressed. Housekeepers are expressed, insofar it is a good idea to start the search for a good loading control there. But it is also utmost important that the loading control is not affected by and experimental difference (like cell types, times, treatments...). And this has to be checked for potential loading controls.

This is a hard job. From now on I will refer to the "experimental difference" as "treatment" (if you are e.g. comparing cell-types, then "cell-type" is the treatment, so to say). The difficulty is that the RT efficiency varies and you may not get absolutely similar amounts of mRNA and cDNA from your different treatments. The only solution is to use a larger panel of candidates (assumed not being related to your treatment!) and compare their treatment-effects. The argument is that is is unlikely that most of these genes will (unexpectedly) be affected by the treatment, so on may identify outliers that show a response that is clearly different to the response of the large rest.

Now even the whole panel (or the "large rest") of candidates may vary like hell, so that no clear pattern is visible. In this case the best you can do is to use as many of these candidates in parallel to normalize the target gene expression (hoping that the different responses of these genes will "average out"). If their pattern is very similar, then you may decide to take only one (or two) of them without much comprising the precision.

Using just one (or two) genes (possibly picked from literature) and seeing that the results of the experiments look nice is NO INDICATION that the loading controls do their job. As long as RT efficiency and sample loading can not be exactly controlled, the only way to check if they really work is to comparatatively analyze a large number of candidates. Well, and if the amount of sample loading can be (almost) perfectly controlled, then I would ask what one needs a loading control for then (measuring the loading controls and using this to normalize the expression of the target gene introduces additional error and uncertainty!).

Conclusions:

* One should avoid the term "housekeeping gene" if one is talking about a gene used as loading control in real-time PCR. A better term is "reference gene".

* Not any housekeeping gene is a valid reference gene, and not any valid reference gene needs to be a housekeeping gene (it is perfectly fine to use a cell-type-speficic gene as reference gene; only its expression must not be influenced by the treatment).

* Finding valid reference genes is difficult, time- and cost intensive. It requires to measure many candidates to be able to confindently identify outliers or common patterns in expression among them. Relying on literature is usually a bad idea, unless the authors really used the same cells and treatments and really did the efforts described above (what, to my experience, is a rare case).

Dear  Prescott Deininger and Dear Adam Brazzle

I work on Breast cancer Cell lines and tissues also I work on circular RNAs of Breast Cancer patients... which one should I use?

thanks

I used GAPDH for differnet human cell lines and tissues, the results was good (expression levels of GAPDH in all samples are nearly  the same)