If you use a 1-step RT-qPCR (RT and PCR in one reaction, with a hot-start polymerase) you only need to normalize the amount of your input RNA. If you do a 2-step RT-qPCR (RT and PCR steps are separated) you normalize the cDNA. It has to be normalized prior to qPCR as this will result in the differential that you are going to measure.

In addition to that, and depending on the purpose of the experiment, use standards/internal controls and triplicates to further control your reaction and verify your findings.

Thank you Peter for your response

I will assume two-step, since this is what I use (and recommend everyone uses).

The main sources of variability are

For 2., you can directly measure this, and you should. This also allows you to check your 260/230 ratio (which is important: low 260/230 ratios usually mean poor cDNA synthesis). You should, wherever possible, use the same amount of input RNA in each cDNA synthesis reaction, because otherwise you're deliberately introducing variability for no reason.

For 3., you can check via bioanalyser or gel electrophoresis, but in my experience this is very rarely a problem. Plus, if you have degradation issues, RNA usually goes from "totally fine" to "useless" pretty quickly, and thus you'll spot you had poor integrity downstream when all the numbers are "N/A". So I would tend to ignore this component unless indicated otherwise (this is also cheaper).

For 4., you really cannot control for this via spectrophotometry, so this is where reference genes come in (see below)

For 5., you presumably do some initial checks to ensure your PCR efficiencies are good and your primer pairs are robust, generating single amplicons and giving linear data over a high dynamic range. If all these are fine, then you can ignore 5. Also, for any given GOI, all samples will have the same PCR conditions, so minor variations essentially cancel out.

Thus the real question is how to dissect 1. from 4.

This is where you use reference genes. You should establish a panel of genes (2-3 is ideal) that do not vary in expression across your experimental conditions: here you can thus assume that variability due to 1. is "zero", and thus any detectable reference gene variation you measure via qPCR is entirely down to 4. (and possibly a bit of 2. and 3. as well).

You then use your reference genes (which have established variation due to 4.) to normalise your GOI data, removing the 4. component of variation from that data. What remains is variation due to 1., the data you are actually interested in.

So:

Measure RNA, do basic QC.

Use the same amount of RNA in each cDNA synthesis reaction.

Use validated reference genes in your qPCR to normalise your expression data.

John Hildyard thank you for the elaborate input. What about the qPCR reaction, should the amount of cDNA be the same for each sample?

John Hildyard are you able to give range of 260/230 ratio values that are indicative of good cDNA synthesis? Can you give reason(s) why more than one reference gene should be used? Do you know of any ideal housekeeping genes to use?

Thank you

>should the amount of cDNA be the same for each sample

...Yes? Remember, you cannot easily quantify cDNA (nanodrop won't tell you anything, because your cDNA reaction still contains all the primers, unincorporated dNTPs and probably quite a lot of the original RNA, all of which absorb at 260). The only thing you can do is assume cDNA synthesis is essentially 1:1, so 800ng RNA becomes 800ng cDNA. Assuming you start with the same amount of RNA in each cDNA synthesis reaction, you should therefore add the same amount (as VOLUME) of each cDNA to your PCRs.

Oh, and before I forget: DILUTE YOUR cDNA. Always dilute it. 1/10-1/20 is fine. cDNA reaction buffer can be quite toxic to PCR, and also undiluted cDNA contains far more template molecules than you need (which can also be inhibitory), so you always need to dilute it.

I use 8ng per well (so from an 1600ng cDNA reaction in 20ul, I dilute it 1/20 to give 4ng.ul-1, and I add 2ul of this diluted stock to each well). This is fine even for low abundance transcripts.

And yes, I add 2ul of cDNA for every sample: I never change the volume of cDNA I add.

>are you able to give range of 260/230 ratio values that are indicative of good cDNA synthesis?

The ideal is anything over 2.0, but you may be able to get away with lower values. The kit I use seems to be happy with anything >1.7, so that's my benchmark. If your 260/230 is below 1.0, you will likely have problems, though remember it's a ratio: if you have very low [RNA], you might have a poor 260/230 but will still be able to make cDNA (albeit not a lot, because you don't have much RNA).

>Can you give reason(s) why more than one reference gene should be used?

Read the MIQE guidelines: https://pubmed.ncbi.nlm.nih.gov/19246619/

One reference gene leaves you very vulnerable to stochastic variation (gene expression sometimes does weird stuff, even with stable genes, and add in freak variations in PCR efficiency or cDNA synthesis and this compounds). Using two reference genes minimises this, and three reduces risk massively. Also, in some cases there really is no truly stable gene, but there might be two that vary in consistently opposite directions: these two together will be a better reference than any other gene alone.

Always aim for two or three.

>Do you know of any ideal housekeeping genes to use?

Impossibly to answer given I don't even know what your sample is prepared from. Prokaryotic? Eukaryotic? Plant? Animal? Tissue? Cell culture?

Even then, you should work on the principle of "no universal reference gene exists", and instead you should determine a suitable panel of reference genes empirically for your specific comparative scenario. This is publishable, by the way, so it's not only not wasted effort, it's also valuable information.

John Hildyard Thank you for the succinct answers to each of my questions. Much appreciation!

Both RNA and cDNA normalization can be used for qRT-PCR, but the choice of normalization method depends on the research question, the nature of the samples, and the experimental design.

Normalization at the RNA stage involves quantifying the RNA input by measuring the expression level of one or more reference genes in the RNA sample before reverse transcription. The advantage of RNA normalization is that it can control for variations in RNA quality, quantity, and integrity. However, RNA normalization assumes that the reverse transcription efficiency is uniform across all samples, which may not always be the case.

Normalization at the cDNA stage involves quantifying the cDNA input by measuring the expression level of one or more reference genes in the cDNA sample after reverse transcription. The advantage of cDNA normalization is that it can control for variations in reverse transcription efficiency, which can be affected by factors such as RNA quality, quantity, and integrity. However, cDNA normalization assumes that the RNA input is uniform across all samples, which may not always be the case.

Normalization at both the RNA and cDNA stages is recommended for accurate and reliable qRT-PCR analysis. This is because normalizing at both stages can control for variations in both RNA input and reverse transcription efficiency, providing a more accurate and robust measurement of gene expression.

These video playlists might be helpful to you:

https://www.youtube.com/playlist?list=PLzBiAPKi8vONSqjcweFWHMRy13y3484bE

https://www.youtube.com/playlist?list=PLzBiAPKi8vOMheKwb0mu1dcZRaeGUFSlw

https://www.youtube.com/playlist?list=PLzBiAPKi8vOPbE7Db9mKwloe34E8rXXEU