Yes, I completely agree with your comment. However, I think that you have already got the list of "gene suspects". I might be wrong but I believe this list was originally obtained from a variety of cancer derived cells or tissues that were not manipulated in any specific way. However, as you're actively manipulating your cellular conditions (LPS is a mitogen, right?), GAPDH (which is involved in energy metabolism) or ACTB (cell structure) may not be very useful in your experiment as endogenous controls since LPS stimulated cells undergo cell division, which is an energy consuming process. Just a thought.  

Stuart,

My personal believe has always been that different experimental condition will require different housekeeping genes for normalization. I know that people prefer to use the set of genes you mentioned as HKG, but ideally you'll have to check all of them and see how stable their expression is between your experimental conditions if you want to use these genes as endogenous controls. Quite frequently genes reported to be stable in one experimental condition show huge coefficient of variation in another experiment. Another suggestion would be to look at microarrays or RNA-Seq data generated from microglia cell lines under conditions similar to your experiment (if such experiments have been previously performed of course) to select the most stable HK genes. Good luck

Thanks Vladimir.

Yes, whatever I start with, I intend to validate their usefulness in *each* experimental condition I use (a variety of signalling molecules).

I'm really just looking for a shortlist of maybe 8 genes. I'll then try to find out whether, for example, LPS affects each one. If there's no obvious reason to rule them out after that, I'll run actual qPCRs and try to determine how reliable they actually are.

I doubt anyone can tell me straight up which ones to use, but perhaps people have experience of which genes are definitely suspect as housekeepers.

Yes, I completely agree with your comment. However, I think that you have already got the list of "gene suspects". I might be wrong but I believe this list was originally obtained from a variety of cancer derived cells or tissues that were not manipulated in any specific way. However, as you're actively manipulating your cellular conditions (LPS is a mitogen, right?), GAPDH (which is involved in energy metabolism) or ACTB (cell structure) may not be very useful in your experiment as endogenous controls since LPS stimulated cells undergo cell division, which is an energy consuming process. Just a thought.  

Yes, LPS can cause mitosis, apoptosis and release of various cytokines.

So, finding genes independent of these responses probably won't be easy...

The substantial morphological changes I see post-LPS addition means I'm ruling out cytoskeletal genes (e.g. B-actin), and as you say, any genes involved in metabolism could well be affected when microglia become activated.

The genes I listed came from various other microglial studies (still frequently using GAPD/GAPDH) and some papers comparing usefulness of housekeeping genes (including across tumours from different tissues).

I currently work with corneal endothelial cells, but a similar question arose in my study. What gene is stably transcribed in ex vivo, primary and cell lines of endothelial cells? I found my answer by combining this helpful review (link provided) with my own transcriptome data. I was fortunate to have the transcriptome data as I could scan for those housekeeping genes that showed the least variation between the different endothelial cell groups, not everyone is that lucky. I chose RAB7, but this gene may not be relevant to your work.

I do have transcriptome data for mouse primary microglia that I treated with IL-6 (not LPS, unfortuntely) over two time points. If you are interested I will check my data for your housekeeping genes of interest and report to you the variability in their transcription between the time points. Let me know.

By the way, during my time in the glia world I used GAPDH and 18S as controls, but that was by default and prior to my appreciating that other genes may have been better.

Best of luck.

http://www.ncbi.nlm.nih.gov/pubmed/23810203

Thanks Ricardo. I have that paper already, and it looks useful.

I very much appreciate the offer to check variability of (even some) my shortlisted genes!

I found a microarray dataset for LPS-treated (rat) microglia, and combined it with a steroid-treated dataset of mine. I've shortlisted some genes from this:

Rpl22, Ube2i, Rpl32, Canx, St13, Gosr2, Hyal2, Usp14, Mt-co2, Atp6ap2, Gusb, Gapdh (frequently criticised, but it actually held up well here; I also want to see how it compares to the other candidates)

What do you think about Mt-co2 (Cox2), *mitochondrial* cytochrome c oxidase II as a housekeeping gene?

Rab7 looks pretty stable with LPS (1.03 fold change), dexamethasone (1.04) and glioma-conditioned medium (0.94), but responded to IFN-gamma (0.10). I might add it in to my selection, though.

Much appreciate your help with this. New project and techniques for me, and I want to get it correct from the start.

Seems that you have a great list of genes to investigate. Depending on the time and resources you have available you can narrow it down to three and perform your qPCR analysis with those. When possible it is best to use multiple housekeeping genes to show reviewers of your work that you have performed adequate controls. You can get away with just one gene and I think in your instance where you are validating prior to the main experiment you will have the optimization data to prove it.

Regarding Cox2,  I haven't used this one before, but having a look at my data I find that at 2hr IL6 tx I get a 0.92 change compared with NT and at 12hr I get a 0.97 change. If I didn't have any other options I might consider using Cox2.

Best of luck.

Stuart, I wanted to add one other comment. When considering your choice for HKG selection keep in mind that for accurate estimation of relative abundance you want the HKG and the TG expressions to be similar. Another point to keep in mind is to make sure that PCR efficiencies for the TGs and the HKGs are at best equal or at worst quite similar. To check PCR efficiency we typically use the LinRegPCR software (Ramakers et al. (2003). Assumption-free analysis of quantitative real-time polymerase chain reaction (PCR) data. Neurosci Lett. 339, 62-66). If PCR efficiencies are not equal you could try to correct for that in your formula, but the results are not as good. Good luck  

Thanks for that link Vladimir. I'm aware of the principle of relative abundance, now I've got a lot of reading and planning ahead of me to make sure I get it right...