I was in the similar situation(with microarrays) and do not have a good solution. In our case cells strongly decreased total RNA content over time. So when we normalise the arrays together we can see some transcripts going up, while they are most probably just less decreased than others. At the end we can only discuss the relative changes of the transcripts composition. I would love to know how other people approach this problem.

Is the reduction or increase in RNA content mostly uniform across all genes (i.e., is it due to activity of something in the transcription machinery)? If so, maybe you could normalize to cell number first and then use a panel of 4 or 5 (maybe more?) reference genes to determine internal normalization.

check this one it might help, "Avoiding Pitfalls of Internal Controls: Validation of

Reference Genes for Analysis by qRT-PCR and Western

Blot throughout Rat Retinal Development

Maurı´cio Rocha-Martins1, Brian Njaine2, Mariana S. Silveira1*"

For normalization at the qRT-PCR, one could add a known quantity of DNA fragment that can be simultaneously amplified and used for normalization. Other option is transfect known quantity of a housekeeping gene plasmid and use that as internal control. Preferably optimize the first option for better consistency. Good luck!

RNA spike ins should be used to answer this question.

Please see "Revisiting Global Gene Expression Analysis", Lovén et al., Cell 2012

Isn't the idea of normalization to account for "small" differences between samples that may be due to slightly different amounts of starting amounts of RNA (it may be hard to get 100% accurate RNA measurements in the first place), cDNA synthesis etc etc. To control for such "natural variations" one uses normalizers. The prerequisite for this procedure is the assumption that the cells you compare have similar amounts of total RNA and mRNA.

In your case it may depend whether ribosomal RNAs are more affected than mRNAs because you basically just measure the content of ribosomal RNA when quantifying your total RNA. This may be irrelevant if all RNA classes are roughly affected at similar levels.

If there is a global reduction per cell then why bother? Isn't that the phenotype you should document?

One answer to this situation is to normalise using an average expression level of all the genes in the assay for each cell type. This allows you to look at the changes in relative levels of your target gene between each cell type regardless of the total RNA or the total mRNA in each cell. Of course this solution is easier to work in Q-RTPCR than on arrays.

Thanks all for the comments.... I am however unsure if taking equal amounts or NA would solve the issue as there is increased transcription in once condition (of several RNAs and not just ribosomal RNAs). I am using different controls for now and will try to normalize..... But I do think that for my experiment, using a couple of spike in controls should be helpful as suggested by Tess..

I do welcome more suggestions and comments in case someone has something interesting and different to suggest..